CN105374033A - SAR image segmentation method based on ridgelet deconvolution network and sparse classification - Google Patents
SAR image segmentation method based on ridgelet deconvolution network and sparse classification Download PDFInfo
- Publication number
- CN105374033A CN105374033A CN201510675676.5A CN201510675676A CN105374033A CN 105374033 A CN105374033 A CN 105374033A CN 201510675676 A CN201510675676 A CN 201510675676A CN 105374033 A CN105374033 A CN 105374033A
- Authority
- CN
- China
- Prior art keywords
- sketch
- region
- sar image
- pixel
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000003709 image segmentation Methods 0.000 title claims abstract description 22
- 230000002776 aggregation Effects 0.000 claims abstract description 87
- 238000004220 aggregation Methods 0.000 claims abstract description 87
- 230000011218 segmentation Effects 0.000 claims abstract description 22
- 230000005428 wave function Effects 0.000 claims description 31
- 230000004044 response Effects 0.000 claims description 30
- 238000003475 lamination Methods 0.000 claims description 17
- 238000011524 similarity measure Methods 0.000 claims description 14
- 230000006870 function Effects 0.000 claims description 12
- 238000010586 diagram Methods 0.000 claims description 10
- VMXUWOKSQNHOCA-UKTHLTGXSA-N ranitidine Chemical compound [O-][N+](=O)\C=C(/NC)NCCSCC1=CC=C(CN(C)C)O1 VMXUWOKSQNHOCA-UKTHLTGXSA-N 0.000 claims description 9
- 238000006073 displacement reaction Methods 0.000 claims description 4
- 238000005457 optimization Methods 0.000 claims description 4
- 238000005070 sampling Methods 0.000 claims description 4
- 230000008859 change Effects 0.000 claims description 3
- HUTDUHSNJYTCAR-UHFFFAOYSA-N ancymidol Chemical compound C1=CC(OC)=CC=C1C(O)(C=1C=NC=NC=1)C1CC1 HUTDUHSNJYTCAR-UHFFFAOYSA-N 0.000 claims description 2
- 238000004364 calculation method Methods 0.000 claims description 2
- 238000011478 gradient descent method Methods 0.000 claims description 2
- 230000001629 suppression Effects 0.000 claims description 2
- 230000000694 effects Effects 0.000 abstract description 3
- 238000001514 detection method Methods 0.000 abstract description 2
- 239000000284 extract Substances 0.000 description 7
- 238000000605 extraction Methods 0.000 description 7
- 238000005516 engineering process Methods 0.000 description 6
- 238000009826 distribution Methods 0.000 description 5
- 238000005192 partition Methods 0.000 description 4
- 238000004458 analytical method Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000008569 process Effects 0.000 description 2
- 239000003086 colorant Substances 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
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
- G06T2207/10044—Radar image
Landscapes
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention discloses a SAR image segmentation method based on a ridgelet deconvolution network and sparse classification and mainly solves a problem that image features are extracted in virtue of human experience in the prior art. The method comprises following steps of: (1) sketching a synthetic aperture radar (SAR) image, (2) dividing the SAR image into different semantic regions; (3) training a ridgelet deconvolution network (RDN) of aggregation regions and a RDN of homogeneous regions; (4) merging similar aggregation regions; (5) merging similar homogeneous regions; (6) based on a watershed method, segmenting a structural region acquired in the step (2), and (7) acquiring the segmented SAR image. The SAR image segmentation method achieves good region consistency, improves the segmentation effect of the SAR image segmentation, and can be used for target detection and identification.
Description
Technical field
The invention belongs to technical field of image processing, further relate to a kind of synthetic-aperture radar (SyntheticApertureRadar, the SAR) image partition method based on ridge ripple deconvolution network and sparse classification in target identification technology field.The present invention can the zones of different of Technologies Against Synthetic Aperture Radar SAR image be split exactly, and can be used for the object detection and recognition of follow-up synthetic-aperture radar SAR image.
Background technology
The segmentation of synthetic-aperture radar SAR image refers to, according to features such as gray scale, texture, structure, aggregations, synthetic-aperture radar SAR image is divided into several mutually disjoint regions, and make these features present similarity in the same area, and between zones of different, present the process of obvious otherness.The object of synthetic-aperture radar SAR image segmentation is the representation simplifying or change image, image is easier to understand and analyzes.The segmentation of synthetic-aperture radar SAR image is the basis of image understanding and decipher, and the quality of segmentation quality directly affects follow-up analysis, identification etc.Usually, it is more accurate to split, and identifies more successful.
Existing synthetic-aperture radar SAR image segmentation method is mainly divided into the method for feature based and the method for Corpus--based Method model.The feature that the method for feature based mainly extracts some synthetic-aperture radar SAR image is split, such as textural characteristics, limit feature and composite character etc.The mode of synthetic-aperture radar SAR image segmentation problem probability is expressed by the method for Corpus--based Method model, is the distribution of some experiences by the feature interpretation of image, such as Nakagami distribution, Gamma distribution, K distribution, G distribution etc.
Linda, Xu Xin, Pan Xuefeng, propose a kind of new MSTARSAR image partition method in the paper " a kind of new MSTARSAR image partition method, 2014,39 (11) " that Haitao Zhang is delivered at it.First the method carries out the feature extraction of attribute scattering center according to the scattering mechanism of atural object, structure attribute scattering center proper vector, then use markov random file to carry out adjacent city, space relationship description in conjunction with attribute scattering center feature to MSTARSAR image, finally use the energy-optimised algorithm of label cost to obtain final segmentation result.The weak point that the method exists is, it is artificial extraction that the method Technologies Against Synthetic Aperture Radar SAR image carries out splitting used feature, artificial selected characteristic is the method for a professional knowledge of requiring great effort very much, need, the feature that can have chosen is to a great extent by experience and fortune, and the quality of the feature therefore manually chosen often becomes the bottleneck of whole system performance.
A kind of SAR image segmentation method based on degree of depth own coding and areal map is disclosed in patent that Xian Electronics Science and Technology University applies at it " SAR image segmentation method based on degree of depth own coding and areal map " (number of patent application 201410751944.2, publication number CN104392456A).The method obtains the areal map divided according to the sketch map of synthetic-aperture radar SAR image, areal map is mapped to former figure and obtains gathering, homogeneous and structural region; Respectively to gathering, homogenous region different degree of depth own coding devices training, obtain the feature of gathering and each point in homogenous region; Build dictionary to gathering and homogenous region respectively, each point is projected to corresponding dictionary and converges out the provincial characteristics of all subregion, carries out cluster respectively to the subregion feature in two class regions; Under sketch line segment instructs, use super-pixel to merge to structural region to split; Merge each region segmentation result and complete SAR image segmentation.The weak point that the method exists is, the degree of depth own coding device of automatic extraction characteristics of image used be input as one-dimensional vector, destroy the spatial structure characteristic of image, therefore, the essential characteristic of image can not be extracted, reduce the precision of SAR image segmentation.
Summary of the invention
The object of the invention is to the deficiency overcoming above-mentioned prior art, propose a kind of (SyntheticApertureRadar, SAR) image partition method based on ridge ripple deconvolution neural network and sparse classification.The method is based on ridge ripple deconvolution neural network, not only can the feature of automatic learning sample, break through the bottleneck of artificial extraction feature, and the spatial relationship that can learn in synthetic-aperture radar SAR image between pixel, extract the essential characteristic of synthetic-aperture radar SAR image, improve the segmentation effect of synthetic-aperture radar SAR image.
For achieving the above object, specific implementation step of the present invention comprises as follows:
(1) sketch SAR image:
To the synthetic-aperture radar SAR image sketch of input, obtain the sketch map of synthetic-aperture radar SAR image;
(2) dividing SAR image is different semantic regions:
Utilize ray completion areal map, synthetic-aperture radar SAR image is divided into aggregation zone, homogenous region and structural region;
(3) ridge ripple deconvolution network RDN is trained respectively to aggregation zone and homogenous region:
(3a) one 4 layers ridge ripple deconvolution network RDN are constructed;
(3b) utilize ridge wave function, respectively initialization is carried out to the bank of filters of 3 warp laminations in the 4 layers of ridge ripple deconvolution network RDN constructed;
(3c) to spatially each aggregation zone disconnected and spatially each homogenous region disconnected, train one 4 layers ridge ripple deconvolution network RDN respectively, obtain the optimal value of ridge ripple deconvolution network median filter group;
(4) similar aggregation zone is merged:
(4a) with each aggregation zone disconnected on the optimal value representation space of the spatially bank of filters of the last one deck of disconnected each aggregation zone training gained ridge ripple deconvolution network;
(4b) adopt the method for sparse classification, calculate and represent spatially the estimating of architectural feature similarity between each aggregation zone disconnected;
(4c) aggregation zone architectural feature similarity measure is greater than the corresponding region of its threshold tau as similar aggregation zone, merge all similar aggregation zones, wherein, τ represents the threshold value of aggregation zone architectural feature similarity measure, the span of τ is τ ∈ [0,1];
(5) similar homogenous region is merged:
(5a) with each homogenous region disconnected on the optimal value representation space of the spatially bank of filters of the last one deck of disconnected each homogenous region training gained ridge ripple deconvolution network;
(5b) adopt the method for sparse classification, calculate and represent spatially the estimating of architectural feature similarity between each homogenous region disconnected;
(5c) homogenous region architectural feature similarity measure is greater than the corresponding region of its threshold value σ as similar homogenous region, merge all similar homogenous region, wherein, σ represents the threshold value of homogenous region architectural feature similarity measure, the span of σ is σ ∈ [0,1];
(6) structural region is split:
The structural region that step (2) obtains is split, obtains the segmentation result of structural region;
(7) SAR image after splitting is obtained:
The homogenous region that the aggregation zone utilizing step (4) to obtain and step (5) obtain, and the structural region that step (6) obtains, obtain the synthetic-aperture radar SAR image after splitting.
The present invention compared with prior art has the following advantages:
First, because the present invention constructs one 4 layers ridge ripple deconvolution network RDN, overcome prior art Technologies Against Synthetic Aperture Radar SAR image to carry out splitting the shortcoming that used feature is artificial extraction, make to adopt the present invention automatically can extract the feature of image, more time saving and energy saving than artificial extraction feature, and the characteristics of image automatically extracted is more accurate than the feature of artificial extraction.
Second, because the present invention utilizes ridge wave function, respectively initialization is carried out to the bank of filters of 3 warp laminations in the 4 layers of ridge ripple deconvolution network RDN constructed, overcome the shortcoming that the architectural characteristic of image do not paid close attention to by degree of depth own coding device that prior art extracts characteristics of image automatically, make to adopt the present invention automatically can extract the structural specificity feature of image, improve the precision of SAR image segmentation.
3rd, due to the present invention to spatially each aggregation zone disconnected and spatially each homogenous region disconnected train one 4 layers ridge ripple deconvolution network RDN respectively, overcome the shortcoming that the spatial relationship in image between pixel do not paid close attention to by degree of depth own coding device that prior art extracts characteristics of image automatically, make to adopt the present invention automatically can extract the essential characteristic of image, therefore, better region segmentation consistance is obtained.
Accompanying drawing explanation
Fig. 1 is process flow diagram of the present invention;
To be Ku wave band resolution be Fig. 2 (a) that the synthetic-aperture radar SAR image Piperiver of 1 meter schemes;
Fig. 2 (b) is the areal map that Piperiver figure is corresponding;
Fig. 2 (c) is Piperiver figure aggregation zone segmentation result figure;
Fig. 2 (d) is Piperiver figure homogenous region segmentation result figure;
Fig. 2 (e) is the final segmentation result figure of Piperiver image.
Embodiment
Below in conjunction with accompanying drawing, the present invention will be further described.
With reference to accompanying drawing 1, concrete steps of the present invention are as follows.
Step 1, sketch is combined to aperture radar SAR image.
Input synthetic-aperture radar SAR image, by its sketch, obtains the sketch map of synthetic-aperture radar SAR image.
The synthetic-aperture radar SAR image sketch model that the present invention uses is the model proposed the people such as Jie-Wu were published on IEEETransactionsonGeoscienceandRemoteSensing magazine article " LocalmaximalhomogenousregionsearchforSARspecklereduction withsketch-basedgeometricalkernelfunction " in 2014 in.
There is the limit of different directions and yardstick, line template, and utilize the direction of template and dimensional information structural anisotropy Gaussian function to calculate the weighting coefficient of every bit in this template, its mesoscale number value is 3 ~ 5, and direction number value is 18.
According to the following formula, the average of the corresponding pixel in synthetic-aperture radar SAR image of calculation template zones of different and variance:
Wherein, μ represents the average of the corresponding pixel in synthetic-aperture radar SAR image of region Ω, and Ω represents a region in template, and g represents the position of a pixel in the Ω of region, and ∈ represents and belongs to symbol, and ∑ represents sum operation, w
grepresent the weight coefficient at g place, position in the Ω of region, w
gspan be w
g∈ [0,1], A
grepresent g corresponding pixel value in synthetic-aperture radar SAR image in position in the Ω of region, ν represents the variance of the corresponding pixel in synthetic-aperture radar SAR image of region Ω.
According to the following formula, the response of each pixel comparison value operator in synthetic-aperture radar SAR image is calculated:
Wherein, R represents the response of each pixel comparison value operator in synthetic-aperture radar SAR image, and min{} represents operation of minimizing, a and b represents the numbering of any two zoness of different in template respectively, μ
aand μ
brepresent the average of respective pixel in region a and region b and synthetic-aperture radar SAR image respectively.
According to the following formula, each pixel is calculated in synthetic-aperture radar SAR image to the response of correlativity operator:
Wherein, C represents that in synthetic-aperture radar SAR image, each pixel is to the response of correlativity operator, a and b represents the numbering of any two zoness of different in template respectively, ν
aand ν
brepresent the variance of the corresponding pixel in synthetic-aperture radar SAR image of region a and region b respectively, μ
aand μ
brepresent the average of respective pixel in region a and region b and synthetic-aperture radar SAR image respectively,
represent square root functions.
According to the following formula, to merge in the response of pixel comparison value operator in synthetic-aperture radar SAR image and synthetic-aperture radar SAR image pixel to the response of correlativity operator, to calculate in synthetic-aperture radar SAR image each pixel to the response of each template:
Wherein, F represents that in synthetic-aperture radar SAR image, each pixel is to the response of each template, R and C to represent in synthetic-aperture radar SAR image that in pixel comparison value operator and synthetic-aperture radar SAR image, pixel is to the response of correlativity operator respectively,
represent square root functions.
Selection has the template of template as pixel in synthetic-aperture radar SAR image of maximum response, and using the intensity of maximum response as this pixel, to the direction of direction as this pixel of the template of maximum response be had, obtain sideline response diagram and the directional diagram of synthetic-aperture radar SAR image.
Utilize the template selected by each pixel in synthetic-aperture radar SAR image, obtain the gradient map of synthetic-aperture radar SAR image.
According to the following formula, the sideline response diagram that will normalize to [0,1] and the gradient map normalizing to [0,1] merge, and obtain intensity map:
Wherein, I represents the intensity level in intensity map, and x represents the value in the response diagram of sideline, and y represents the value in gradient map.
Adopt non-maxima suppression method, intensity map is detected, obtains suggestion sketch.
Choose the pixel in suggestion sketch with maximum intensity, the pixel be communicated with is connected to form suggestion line segment, obtains suggestion sketch map in suggestion sketch with the pixel of this maximum intensity.
According to the following formula, the code length gain CLG of sketch line in suggestion sketch map is calculated:
Wherein, CLG represents the code length gain of sketch line in suggestion sketch map, and ∑ represents sum operation, and m represents the number of pixel in current sketch line neighborhood, and t represents the numbering of pixel in current sketch line neighborhood, A
trepresent the observed reading of t pixel in current sketch line neighborhood, A
t, 0represent under current sketch line can not represent the hypothesis of structural information, the estimated value of t pixel in this sketch line neighborhood, ln () expression take e as the log operations at the end, A
t, 1represent under current sketch line can represent the hypothesis of structural information, the estimated value of t pixel in this sketch line neighborhood.
The span of setting threshold value T, T is 5 ~ 50, selects the suggestion sketch line of CLG>T as the sketch line in final sketch map, obtains the sketch map that input synthetic-aperture radar SAR image is corresponding.
Step 2, dividing SAR image is different semantic regions.
According to the concentration class of sketch line segment in the sketch map of synthetic-aperture radar SAR image, sketch line is divided into the gathering sketch line representing and assemble atural object and the sketch line representing border, line target and isolated target;
According to the statistics with histogram of sketch line segment concentration class, choose concentration class and equal the sketch line segment of optimum concentration class as seed line-segment sets { E
k, k=1,2 ..., m}, wherein E
krepresent arbitrary sketch line segment in seed line-segment sets, k is the label of arbitrary sketch line segment in seed line-segment sets, and m is the total number of line segment, and { } represents set operation;
If seed line-segment sets { E
k, k=1,2 ..., the sketch line segment E in m}
kbe not added into certain line segment aggregate, then with sketch line segment E
knew line segment aggregate is solved for basic point recurrence;
Actionradius is the circular primitive in the interval upper bound of optimum concentration class, first expands to the line segment in line segment aggregate, corrodes, sketch map obtains the aggregation zone in units of sketch point to the line segment aggregate ecto-entad after expanding;
Calculate the length of each root sketch line in the sketch line representing border, line target and isolated target, from long to short these sketch lines are sorted according to length, obtain the sketch line set after sorting;
The initial value of counter is set to 1, the value of the threshold value E of counter α is set to 21;
Judge whether counter is less than threshold value, if so, then perform the 8th step, otherwise, perform the 13rd step;
3 Seed Points selected by α root sketch line after sequence, is divided into 4 grades to divide line segment on sketch line with these 3 Seed Points, if certain Seed Points is the end points of sketch line just, then this Seed Points is moved to the midpoint of place sketch line segment;
In α root sketch line both sides, with each Seed Points on α root sketch line for starting point, 180 directions being 1 ~ 180 degree along the angle between sketch line stretch out, during extension is run into and represents border, the sketch line of line target and isolated target, the border of aggregation zone, any one in three kinds, the border situation of the closed region that other sketch line completion obtains, stop extending, obtaining take Seed Points as the ray of starting point, each Seed Points respectively produces 180 articles of rays in α root sketch line both sides, according to the size at ray and sketch wire clamp angle, these rays produced are sorted,
Calculate the length of all rays, first ray bunch is generated with Article 1 ray, for remaining 179 rays, judge that the length ratio of the last bar ray that the length of wherein each ray is adjacent is whether between 1.25 ~ 1.5, if then this ray to be added the ray bunch at its last bar ray place, otherwise, generate a new ray bunch with this ray;
The ray bunch of 5 is less than and the ray of length sudden change is revised to comprising number of rays;
Connect the end points of current sketch line and the ray terminal with its arest neighbors on locus, obtain the ray closed level of Seed Points;
Merge the ray closed level of 3 Seed Points, obtain the ray closed level of sketch line, and utilize this ray closed level to obtain the sketch line of completion, obtain the closed region of current sketch line, the value of counter α is increased by 1, perform the 7th step;
To expression border, line target and the sketch line of isolated target and the sketch line of its completion, centered by each sketch point of each sketch line, construct the geometry window acquisition structural region that size is 5 × 5;
Using the part that removes in sketch map beyond aggregation zone and structural region as can not sketch region;
By the aggregation zone in sketch map, structural region with can not correspond in synthetic-aperture radar SAR image in sketch region, obtain the aggregation zone of synthetic-aperture radar SAR image, structural region and homogenous region.
Step 3, trains ridge ripple deconvolution network RDN respectively to aggregation zone and homogenous region.
1st step, constructs one 4 layers ridge ripple deconvolution network RDN.
Input layer is set to by the 1st layer of ridge ripple deconvolution network; Be set to warp lamination by the 2nd layer of ridge ripple deconvolution network, warp lamination comprises the wave filter of 97 × 7 sizes and the characteristic pattern of 9 37 × 37 sizes, and the wave filter of 97 × 7 sizes forms a bank of filters; Be set to warp lamination by the 3rd layer of ridge ripple deconvolution network, warp lamination comprises the wave filter of 45 7 × 7 sizes and the characteristic pattern of 45 43 × 43 sizes, and the wave filter of 45 7 × 7 sizes forms a bank of filters; Be set to warp lamination by the 4th layer of ridge ripple deconvolution network, warp lamination comprises the wave filter of 100 7 × 7 sizes and the characteristic pattern of 100 49 × 49 sizes, and the wave filter of 100 7 × 7 sizes forms a bank of filters.
2nd step, utilizes ridge wave function, carries out initialization respectively to the bank of filters of 3 warp laminations in the 4 layers of ridge ripple deconvolution network RDN constructed.
Carry out discretize to the parameter of continuous ridge wave function, obtain the discretize parameter of ridge wave function, described continuous ridge wave function formula is as follows:
Wherein, A represents continuous ridge wave function, and a represents the scale parameter of continuous ridge wave function, the span of a be a ∈ (0,3], discretize is spaced apart 1, ∈ and represents and belong to symbol, and ψ () represents wavelet function, x
1and x
2represent horizontal ordinate and the ordinate of pixel in the wave filter in deconvolution layer bank of filters respectively, θ represents the direction parameter of continuous ridge wave function, the span of θ is θ ∈ [0, π), discretize is spaced apart π/18, b represents the displacement parameter of continuous ridge wave function, when direction parameter θ is at θ ∈ [0, pi/2) in scope during value, the span of b is b ∈ [0, n × (sin θ+cos θ)], when direction parameter θ is at θ ∈ [pi/2, π) in scope during value, the span of b is b ∈ [n × cos θ, n × sin θ], n represents the threshold parameter of displacement parameter b, the span of n is n ∈ (0, 1], sin represents sine function, cos represents cosine function, the discretize of b is spaced apart 1.
Appoint the ridge wave function parameter value getting 9 groups of discretizes, utilize the continuous ridge wave function formula in the 1st step, initialization is carried out to the bank of filters of the 2nd layer, ridge ripple deconvolution network.
Appoint the ridge wave function parameter value getting 45 groups of discretizes, utilize the continuous ridge wave function formula in the 1st step, initialization is carried out to the bank of filters of the 3rd layer, ridge ripple deconvolution network.
Appoint the ridge wave function parameter value getting 100 groups of discretizes, utilize the continuous ridge wave function formula in the 1st step, initialization is carried out to the 4th layer of bank of filters of ridge ripple deconvolution network.
3rd step, to spatially each aggregation zone disconnected and spatially each homogenous region disconnected train one 4 layers ridge ripple deconvolution network RDN respectively, obtain the optimal value of ridge ripple deconvolution network median filter group.
The method of described training deconvolution network, to be published in the article " DeconvolutionalNetworks " on meeting ComputerVisionandPatternRecognition in 2010 see people such as MatthewD.Zeiler, this is a kind of method extracting characteristics of image without supervision level.
Carry out intensive sliding window sampling to aggregation zone and homogenous region respectively, sample window size is respectively 31 × 31 pixels and 17 × 17 pixels, obtains the sample of aggregation zone or homogenous region sampling.
Respectively by the sample of aggregation zone and homogenous region, be input in 4 layers of ridge ripple deconvolution network RDN.
In fixing ridge ripple deconvolution network, the value of characteristic pattern and bank of filters, by solving an one-dimensional optimization problem, obtains the optimal value of auxiliary variable in ridge ripple deconvolution network.
In fixing ridge ripple deconvolution network, the value of auxiliary variable and bank of filters, by solving a linear system optimization problem, obtains the optimal value of characteristic pattern in ridge ripple deconvolution network.
In fixing ridge ripple deconvolution network, the value of characteristic pattern and auxiliary variable, by using gradient descent method, obtains the optimal value of ridge ripple deconvolution network median filter group.
Step 4, merges similar aggregation zone.
1st step, represents each aggregation zone by the optimal value of the bank of filters of the last one deck of each aggregation zone training gained ridge ripple deconvolution network.
2nd step, by the method for sparse classification, obtains representing architectural feature similarity measure between each aggregation zone.
Appoint and get an aggregation zone and be set to projection aggregation zone, appoint and get an aggregation zone different with the aggregation zone that projects and be set to and be projected aggregation zone.
By representing that the bank of filters of the last one deck of ridge ripple deconvolution network of projection aggregation zone is set to the projection aggregation zone bank of filters comprising 100 wave filters, the bank of filters of the last one deck of ridge ripple deconvolution network being projected aggregation zone is set to comprise 100 wave filters be projected aggregation zone bank of filters.
Described projection formula is as follows:
Wherein, d represents that projection aggregation zone wave filter or projection homogenous region wave filter project to the projection value being projected aggregation zone wave filter or being projected homogenous region wave filter, and the span of d is d ∈ [0,1], and ∈ represents and belongs to symbol, F
1represent projection aggregation zone wave filter or projection homogenous region wave filter, F
2represent and be projected aggregation zone wave filter or be projected homogenous region wave filter, * represents dot product operations, || || represent and ask modulo operation;
3rd step, is greater than the corresponding region of its threshold tau as similar aggregation zone, merges all similar aggregation zones using aggregation zone architectural feature similarity measure, wherein, τ represents the threshold value of aggregation zone architectural feature similarity measure, and the span of τ is τ ∈ [0,1].
Step 5, merges similar homogenous region.
1st step, represents each homogenous region by the optimal value of the bank of filters of the last one deck of regional training gained ridge ripple deconvolution network.
2nd step, by the method for sparse classification, obtains representing architectural feature similarity measure between each homogenous region.
Appoint and get a homogenous region and be set to projection homogenous region, appoint and get one and to be set to the different homogenous region, homogenous region that projects and to be projected homogenous region.
By representing that the bank of filters of the last one deck of ridge ripple deconvolution network of projection homogenous region is set to the projection homogenous region bank of filters comprising 100 wave filters, the bank of filters of the last one deck of ridge ripple deconvolution network being projected homogenous region is set to comprise 100 wave filters be projected homogenous region bank of filters.
Described projection formula is as follows:
Wherein, d represents that projection homogenous region wave filter projects to the projection value being projected homogenous region wave filter, and the span of d is d ∈ [0,1], and ∈ represents and belongs to symbol, F
1represent projection homogenous region wave filter, F
2represent and be projected homogenous region wave filter, * represents dot product operations, || || represent and ask modulo operation;
3rd step, is greater than the corresponding region of its threshold value σ as similar homogenous region, merges all similar homogenous region using homogenous region architectural feature similarity measure, wherein, σ represents the threshold value of homogenous region architectural feature similarity measure, and the span of σ is σ ∈ [0,1].
Step 6, splits structural region.
Adopt watershed algorithm, structural region is divided into super-pixel.
In the sketch map of synthetic-aperture radar SAR image, parallel and that distance is less than 7 pixels two sketch lines are defined as first kind line target sketch line, the super-pixel between first kind line target sketch line is merged, as first kind line target.
In the initial sketch map of synthetic-aperture radar SAR image, sketch line sketch line both sides being belonged to the same area is defined as Equations of The Second Kind line target sketch line, a pixel is respectively expanded as Equations of The Second Kind line target, using other sketch line as the sketch line portraying border in Equations of The Second Kind line target sketch line both sides.
To each super-pixel except the super-pixel that line target and border cover, will to be adjacent and the super-pixel that the difference of gray average is less than 25 merges, until there are not two super-pixel that the adjacent and difference of gray average is less than 25.
By each super-pixel after merging in the 4th step, be merged into minimum with the difference of this super-pixel gray-scale value average respectively and be less than in the homogenous region of 25, obtaining the result after to structural region segmentation.
Step 7, obtains the synthetic-aperture radar SAR image after splitting.
The homogenous region that the aggregation zone utilizing step 4 to obtain and step 5 obtain, and the structural region that step 6 obtains, obtain the synthetic-aperture radar SAR image after splitting.
Below in conjunction with analogous diagram, the present invention will be further described.
1. simulated conditions:
The hardware condition of the present invention's emulation is: window7, CPUPentium (R) 4, and basic frequency is 3.0GHZ; Software platform is: MatlabR2012a; The present invention emulates the synthetic-aperture radar SAR image used: Ku wave band resolution is the Piperiver figure of 1 meter.
2. emulate content:
Emulation experiment of the present invention splits Piperiver figure, Piperiver figure as shown in Fig. 2 (a) derives from the synthetic-aperture radar SAR image that Ku wave band resolution is 1 meter, extracted region is carried out to Fig. 2 (a), obtains the areal map as shown in Fig. 2 (b).
Adopt the aggregation zone of method of the present invention to Piperiver figure to split, obtain the aggregation zone segmentation result figure shown in Fig. 2 (c), wherein the region representation same atural object of same color, the atural object that the region representation of different colours is different.Adopt the homogenous region of method of the present invention to Piperiver figure to split, obtain the segmentation result figure of the homogenous region shown in Fig. 2 (d), the region representation same atural object that wherein color is identical, the atural object that the region representation that color is different is different.The structural region of method of the present invention to Piperiver figure is adopted to split, and structural region segmentation result is merged in the segmentation result of homogenous region, obtain the final synthetic-aperture radar SAR image segmentation result figure as shown in Fig. 2 (e), the region representation same atural object that wherein color is identical, the atural object that the region representation that color is different is different.
3. simulated effect analysis:
Can be seen by the segmentation result of the Piperiver figure shown in above-mentioned Fig. 2 (e), use the inventive method Technologies Against Synthetic Aperture Radar SAR image to carry out splitting the precision that can improve segmentation, the region consistency in segmentation result is better.
Claims (8)
1., based on a SAR image segmentation method for ridge ripple deconvolution network and sparse classification, comprise the steps:
(1) sketch SAR image:
To the synthetic-aperture radar SAR image sketch of input, obtain the sketch map of synthetic-aperture radar SAR image;
(2) dividing SAR image is different semantic regions:
Utilize ray completion areal map, synthetic-aperture radar SAR image is divided into aggregation zone, homogenous region and structural region;
(3) ridge ripple deconvolution network RDN is trained respectively to aggregation zone and homogenous region:
(3a) one 4 layers ridge ripple deconvolution network RDN are constructed;
(3b) utilize ridge wave function, respectively initialization is carried out to the bank of filters of 3 warp laminations in the 4 layers of ridge ripple deconvolution network RDN constructed;
(3c) to spatially each aggregation zone disconnected and spatially each homogenous region disconnected, train one 4 layers ridge ripple deconvolution network RDN respectively, obtain the optimal value of ridge ripple deconvolution network median filter group;
(4) similar aggregation zone is merged:
(4a) with each aggregation zone disconnected on the optimal value representation space of the spatially bank of filters of the last one deck of disconnected each aggregation zone training gained ridge ripple deconvolution network;
(4b) adopt the method for sparse classification, calculate and represent spatially the estimating of architectural feature similarity between each aggregation zone disconnected;
(4c) aggregation zone architectural feature similarity measure is greater than the corresponding region of its threshold tau as similar aggregation zone, merge all similar aggregation zones, wherein, τ represents the threshold value of aggregation zone architectural feature similarity measure, the span of τ is τ ∈ [0,1];
(5) similar homogenous region is merged:
(5a) with each homogenous region disconnected on the optimal value representation space of the spatially bank of filters of the last one deck of disconnected each homogenous region training gained ridge ripple deconvolution network;
(5b) adopt the method for sparse classification, calculate and represent spatially the estimating of architectural feature similarity between each homogenous region disconnected;
(5c) homogenous region architectural feature similarity measure is greater than the corresponding region of its threshold value σ as similar homogenous region, merge all similar homogenous region, wherein, σ represents the threshold value of homogenous region architectural feature similarity measure, the span of σ is σ ∈ [0,1];
(6) structural region is split:
The structural region that step (2) obtains is split, obtains the segmentation result of structural region;
(7) SAR image after splitting is obtained:
The homogenous region that the aggregation zone utilizing step (4) to obtain and step (5) obtain, and the structural region that step (6) obtains, obtain the synthetic-aperture radar SAR image after splitting.
2. the SAR image segmentation method based on ridge ripple deconvolution network and sparse classification according to claim 1, is characterized in that: the concrete steps of step (1) described sketch are as follows:
1st step, there is the limit of different directions and yardstick, line template, and utilizes the direction of template and dimensional information structural anisotropy Gaussian function to calculate the weighting coefficient of every bit in this template, and its mesoscale number value is 3 ~ 5, and direction number value is 18;
2nd step, according to the following formula, the average of the corresponding pixel in synthetic-aperture radar SAR image of calculation template zones of different and variance:
Wherein, μ represents the average of the corresponding pixel in synthetic-aperture radar SAR image of region Ω, and Ω represents a region in template, and g represents the position of a pixel in the Ω of region, and ∈ represents and belongs to symbol, and ∑ represents sum operation, w
grepresent the weight coefficient at g place, position in the Ω of region, w
gspan be w
g∈ [0,1], A
grepresent g corresponding pixel value in synthetic-aperture radar SAR image in position in the Ω of region, ν represents the variance of the corresponding pixel in synthetic-aperture radar SAR image of region Ω;
3rd step, according to the following formula, calculates the response of each pixel comparison value operator in synthetic-aperture radar SAR image:
Wherein, R represents the response of each pixel comparison value operator in synthetic-aperture radar SAR image, and min{} represents operation of minimizing, a and b represents the numbering of any two zoness of different in template respectively, μ
aand μ
brepresent the average of respective pixel in region a and region b and synthetic-aperture radar SAR image respectively;
4th step, according to the following formula, in calculating synthetic-aperture radar SAR image, each pixel is to the response of correlativity operator:
Wherein, C represents that in synthetic-aperture radar SAR image, each pixel is to the response of correlativity operator, a and b represents the numbering of any two zoness of different in template respectively, ν
aand ν
brepresent the variance of the corresponding pixel in synthetic-aperture radar SAR image of region a and region b respectively, μ
aand μ
brepresent the average of respective pixel in region a and region b and synthetic-aperture radar SAR image respectively,
represent square root functions;
5th step, according to the following formula, to merge in the response of pixel comparison value operator in synthetic-aperture radar SAR image and synthetic-aperture radar SAR image pixel to the response of correlativity operator, to calculate in synthetic-aperture radar SAR image each pixel to the response of each template:
Wherein, F represents that in synthetic-aperture radar SAR image, each pixel is to the response of each template, R and C to represent in synthetic-aperture radar SAR image that in pixel comparison value operator and synthetic-aperture radar SAR image, pixel is to the response of correlativity operator respectively,
represent square root functions;
6th step, selection has the template of template as pixel in synthetic-aperture radar SAR image of maximum response, and using the intensity of maximum response as this pixel, to the direction of direction as this pixel of the template of maximum response be had, obtain sideline response diagram and the directional diagram of synthetic-aperture radar SAR image;
7th step, utilizes the template selected by each pixel in synthetic-aperture radar SAR image, obtains the gradient map of synthetic-aperture radar SAR image;
8th step, the sideline response diagram that according to the following formula, will normalize to [0,1] and the gradient map normalizing to [0,1] merge, and obtain intensity map:
Wherein, I represents the intensity level in intensity map, and x represents the value in the response diagram of sideline, and y represents the value in gradient map;
9th step, adopts non-maxima suppression method, detects, obtain suggestion sketch to intensity map;
10th step, chooses the pixel in suggestion sketch with maximum intensity, the pixel be communicated with is connected to form suggestion line segment, obtains suggestion sketch map in suggestion sketch with the pixel of this maximum intensity;
11st step, according to the following formula, calculates the code length gain CLG of sketch line in suggestion sketch map:
Wherein, CLG represents the code length gain of sketch line in suggestion sketch map, and ∑ represents sum operation, and m represents the number of pixel in current sketch line neighborhood, and t represents the numbering of pixel in current sketch line neighborhood, A
trepresent the observed reading of t pixel in current sketch line neighborhood, A
t, 0represent under current sketch line can not represent the hypothesis of structural information, the estimated value of t pixel in this sketch line neighborhood, ln () expression take e as the log operations at the end, A
t, 1represent under current sketch line can represent the hypothesis of structural information, the estimated value of t pixel in this sketch line neighborhood;
12nd step, the span of setting threshold value T, T is 5 ~ 50, selects the suggestion sketch line of CLG>T as the sketch line in final sketch map, obtains the sketch map that input synthetic-aperture radar SAR image is corresponding.
3. the SAR image segmentation method based on ridge ripple deconvolution network and sparse classification according to claim 1, is characterized in that: the division SAR image described in step (2) is that the concrete steps in different semantic region are as follows:
1st step, according to the concentration class of sketch line segment in the sketch map of synthetic-aperture radar SAR image, is divided into the gathering sketch line representing and assemble atural object and the sketch line representing border, line target and isolated target by sketch line;
2nd step, according to the statistics with histogram of sketch line segment concentration class, chooses concentration class and equals the sketch line segment of optimum concentration class as seed line-segment sets { E
k, k=1,2 ..., m}, wherein E
krepresent arbitrary sketch line segment in seed line-segment sets, k is the label of arbitrary sketch line segment in seed line-segment sets, and m is the total number of line segment, and { } represents set operation;
3rd step, if seed line-segment sets { E
k, k=1,2 ..., the sketch line segment E in m}
kbe not added into certain line segment aggregate, then with sketch line segment E
knew line segment aggregate is solved for basic point recurrence;
4th step, actionradius is the circular primitive in the interval upper bound of optimum concentration class, first expands to the line segment in line segment aggregate, corrodes, sketch map obtains the aggregation zone in units of sketch point to the line segment aggregate ecto-entad after expanding;
5th step, calculates the length of each root sketch line in the sketch line representing border, line target and isolated target, sorts from long to short according to length to these sketch lines, obtains the sketch line set after sorting;
6th step, is set to 1 by the initial value of counter, and the value of the threshold value E of counter α is set to 21;
7th step, judges whether counter is less than threshold value, if so, then performs the 8th step, otherwise, perform the 13rd step;
8th step, 3 Seed Points is selected by the α root sketch line after sequence, is divided into 4 grades to divide line segment on sketch line, if certain Seed Points is the end points of sketch line just, then this Seed Points is moved to the midpoint of place sketch line segment with these 3 Seed Points;
9th step, in α root sketch line both sides, with each Seed Points on α root sketch line for starting point, 180 directions being 1 ~ 180 degree along the angle between sketch line stretch out, during extension is run into and represents border, the sketch line of line target and isolated target, the border of aggregation zone, any one in three kinds, the border situation of the closed region that other sketch line completion obtains, stop extending, obtaining take Seed Points as the ray of starting point, each Seed Points respectively produces 180 articles of rays in α root sketch line both sides, according to the size at ray and sketch wire clamp angle, these rays produced are sorted,
10th step, calculate the length of all rays, first ray bunch is generated with Article 1 ray, for remaining 179 rays, judge that the length ratio of the last bar ray that the length of wherein each ray is adjacent is whether between 1.25 ~ 1.5, if then this ray to be added the ray bunch at its last bar ray place, otherwise, generate a new ray bunch with this ray;
11st step, is less than the ray bunch of 5 and the ray of length sudden change is revised to comprising number of rays;
12nd step, connects the end points of current sketch line and the ray terminal with its arest neighbors on locus, obtains the ray closed level of Seed Points;
13rd step, merges the ray closed level of 3 Seed Points, obtains the ray closed level of sketch line, and utilize this ray closed level to obtain the sketch line of completion, obtain the closed region of current sketch line, and the value of counter α is increased by 1, performs the 7th step;
14th step, to expression border, line target and the sketch line of isolated target and the sketch line of its completion, constructs the geometry window acquisition structural region that size is 5 × 5 centered by each sketch point of each sketch line;
15th step, using the part that removes in sketch map beyond aggregation zone and structural region as can not sketch region;
16th step, by the aggregation zone in sketch map, structural region with can not correspond in synthetic-aperture radar SAR image in sketch region, obtains the aggregation zone of synthetic-aperture radar SAR image, structural region and homogenous region.
4. the SAR image segmentation method based on ridge ripple deconvolution network and sparse classification according to claim 1, is characterized in that: the concrete steps of one the 4 layers ridge ripple deconvolution network of the structure described in step (3a) are as follows:
1st step, is set to input layer by the 1st layer of ridge ripple deconvolution network;
2nd step, is set to warp lamination by the 2nd layer of ridge ripple deconvolution network, and warp lamination comprises the wave filter of 97 × 7 sizes and the characteristic pattern of 9 37 × 37 sizes, and the wave filter of 97 × 7 sizes forms a bank of filters;
3rd step, is set to warp lamination by the 3rd layer of ridge ripple deconvolution network, and warp lamination comprises the wave filter of 45 7 × 7 sizes and the characteristic pattern of 45 43 × 43 sizes, and the wave filter of 45 7 × 7 sizes forms a bank of filters;
4th step, is set to warp lamination by the 4th layer of ridge ripple deconvolution network, and warp lamination comprises the wave filter of 100 7 × 7 sizes and the characteristic pattern of 100 49 × 49 sizes, and the wave filter of 100 7 × 7 sizes forms a bank of filters.
5. the SAR image segmentation method based on ridge ripple deconvolution network and sparse classification according to claim 1, is characterized in that: described in step (3b) to carry out initialized concrete steps to the bank of filters of 3 warp laminations respectively by ridge wave function as follows:
1st step, carries out discretize to the parameter of continuous ridge wave function, obtains the discretize parameter of ridge wave function, and described continuous ridge wave function formula is as follows:
Wherein, A represents continuous ridge wave function, and a represents the scale parameter of continuous ridge wave function, the span of a be a ∈ (0,3], discretize is spaced apart 1, ∈ and represents and belong to symbol, and ψ () represents wavelet function, x
1and x
2represent horizontal ordinate and the ordinate of pixel in the wave filter in deconvolution layer bank of filters respectively, θ represents the direction parameter of continuous ridge wave function, the span of θ is θ ∈ [0, π), discretize is spaced apart π/18, b represents the displacement parameter of continuous ridge wave function, when direction parameter θ is at θ ∈ [0, pi/2) in scope during value, the span of b is b ∈ [0, n × (sin θ+cos θ)], when direction parameter θ is at θ ∈ [pi/2, π) in scope during value, the span of b is b ∈ [n × cos θ, n × sin θ], n represents the threshold parameter of displacement parameter b, the span of n is n ∈ (0, 1], sin represents sine function, cos represents cosine function, the discretize of b is spaced apart 1,
2nd step, appoints the ridge wave function parameter value getting 9 groups of discretizes, utilizes the continuous ridge wave function in the 1st step, carry out initialization to the bank of filters of the 2nd layer, ridge ripple deconvolution network;
3rd step, appoints the ridge wave function parameter value getting 45 groups of discretizes, utilizes the continuous ridge wave function in the 1st step, carry out initialization to the bank of filters of the 3rd layer, ridge ripple deconvolution network;
4th step, appoints the ridge wave function parameter value getting 100 groups of discretizes, utilizes the continuous ridge wave function in the 1st step, carry out initialization to the 4th layer of bank of filters of ridge ripple deconvolution network.
6. the SAR image segmentation method based on ridge ripple deconvolution network and sparse classification according to claim 1, is characterized in that: the concrete steps of one the 4 layers ridge ripple deconvolution network RDN of the training described in step (3c) are as follows:
1st step, carry out intensive sliding window sampling to aggregation zone and homogenous region respectively, sample window size is respectively 31 × 31 pixels and 17 × 17 pixels, obtains the sample of aggregation zone or homogenous region sampling;
2nd step, respectively by the sample of aggregation zone and homogenous region, is input in 4 layers of ridge ripple deconvolution network RDN;
3rd step, in fixing ridge ripple deconvolution network, the value of characteristic pattern and bank of filters, by solving an one-dimensional optimization problem, obtains the optimal value of auxiliary variable in ridge ripple deconvolution network;
4th step, in fixing ridge ripple deconvolution network, the value of auxiliary variable and bank of filters, by solving a linear system optimization problem, obtains the optimal value of characteristic pattern in ridge ripple deconvolution network;
5th step, in fixing ridge ripple deconvolution network, the value of characteristic pattern and auxiliary variable, by using gradient descent method, obtains the optimal value of ridge ripple deconvolution network median filter group.
7. the SAR image segmentation method based on ridge ripple deconvolution network and sparse classification according to claim 1, is characterized in that: the concrete steps of the method for step (4b) and the sparse classification described in step (5b) are as follows:
1st step, appoint and get an aggregation zone or homogenous region and be set to projection aggregation zone or projection homogenous region, appoint get one with the different aggregation zone of aggregation zone or the homogenous region that projects or the homogenous region of projecting, be set to and be projected aggregation zone or be projected homogenous region;
2nd step, the bank of filters of last one deck of ridge ripple deconvolution network of projection aggregation zone will be represented, be set to the projection aggregation zone bank of filters comprising 100 wave filters, the bank of filters of the last one deck of ridge ripple deconvolution network representing projection homogenous region is set to the projection homogenous region bank of filters comprising 100 wave filters, the bank of filters of the last one deck of ridge ripple deconvolution network being projected aggregation zone is set to comprise 100 wave filters be projected aggregation zone bank of filters, the bank of filters of the last one deck of ridge ripple deconvolution network being projected homogenous region is set to comprise 100 wave filters be projected homogenous region bank of filters,
3rd step, according to projection formula, projects any one wave filter in projection aggregation zone bank of filters to all wave filters be projected in aggregation zone bank of filters, obtains 100 groups of projection values; According to projection formula, any one wave filter in the bank of filters of projection homogenous region is projected to all wave filters be projected in the bank of filters of homogenous region, obtains 100 groups of projection values;
Described projection formula is as follows:
Wherein, d represents that projection aggregation zone wave filter or projection homogenous region wave filter project to the projection value being projected aggregation zone wave filter or being projected homogenous region wave filter, and the span of d is d ∈ [0,1], and ∈ represents and belongs to symbol, F
1represent projection aggregation zone wave filter or projection homogenous region wave filter, F
2represent and be projected aggregation zone wave filter or be projected homogenous region wave filter, * represents dot product operations, || || represent and ask modulo operation;
4th step, is added minimum value of each group of 100 groups of projection values, then divided by 100, obtains estimating of architectural feature similarity between expression two aggregation zones or two homogenous region.
8. the SAR image segmentation method based on ridge ripple deconvolution network and sparse classification according to claim 1, is characterized in that: the concrete steps split structural region described in step (6) are as follows:
1st step, adopts watershed algorithm, structural region is divided into super-pixel;
Parallel and that distance is less than 7 pixels two sketch lines, in the sketch map of synthetic-aperture radar SAR image, are defined as first kind line target sketch line, merge, the super-pixel between first kind line target sketch line as first kind line target by the 2nd step;
3rd step, in the initial sketch map of synthetic-aperture radar SAR image, sketch line sketch line both sides being belonged to the same area is defined as Equations of The Second Kind line target sketch line, a pixel is respectively expanded as Equations of The Second Kind line target, using other sketch line as the sketch line portraying border in Equations of The Second Kind line target sketch line both sides;
4th step, to each super-pixel except the super-pixel that line target and border cover, will to be adjacent and the super-pixel that the difference of gray average is less than 25 merges, until there are not two super-pixel that the adjacent and difference of gray average is less than 25;
5th step, by each super-pixel after merging in the 4th step, is merged into minimum with the difference of this super-pixel gray-scale value average respectively and is less than in the homogenous region of 25, obtains the result after to structural region segmentation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510675676.5A CN105374033B (en) | 2015-10-19 | 2015-10-19 | SAR image segmentation method based on ridge ripple deconvolution network and sparse classification |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510675676.5A CN105374033B (en) | 2015-10-19 | 2015-10-19 | SAR image segmentation method based on ridge ripple deconvolution network and sparse classification |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105374033A true CN105374033A (en) | 2016-03-02 |
| CN105374033B CN105374033B (en) | 2017-11-21 |
Family
ID=55376205
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510675676.5A Expired - Fee Related CN105374033B (en) | 2015-10-19 | 2015-10-19 | SAR image segmentation method based on ridge ripple deconvolution network and sparse classification |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105374033B (en) |
Cited By (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106529570A (en) * | 2016-10-14 | 2017-03-22 | 西安电子科技大学 | Image classification method based on deep ridgelet neural network |
| CN106611423A (en) * | 2016-12-30 | 2017-05-03 | 西安电子科技大学 | SAR image segmentation method based on ridge wave filter and deconvolution structural model |
| CN106683102A (en) * | 2016-12-30 | 2017-05-17 | 西安电子科技大学 | SAR image segmentation method based on ridgelet filters and convolution structure model |
| CN106846322A (en) * | 2016-12-30 | 2017-06-13 | 西安电子科技大学 | Based on the SAR image segmentation method that curve wave filter and convolutional coding structure learn |
| CN106952269A (en) * | 2017-02-24 | 2017-07-14 | 北京航空航天大学 | The reversible video foreground object sequence detection dividing method of neighbour and system |
| CN107292268A (en) * | 2017-06-23 | 2017-10-24 | 西安电子科技大学 | The SAR image semantic segmentation method of quick ridge ripple deconvolution Structure learning model |
| CN107341813A (en) * | 2017-06-15 | 2017-11-10 | 西安电子科技大学 | SAR image segmentation method based on structure learning and sketch characteristic inference network |
| CN108334851A (en) * | 2018-02-12 | 2018-07-27 | 西安电子科技大学 | Based on each to heterogeneous rapid polarization SAR image segmentation method |
| CN108445466A (en) * | 2018-05-04 | 2018-08-24 | 西安电子科技大学 | A kind of radar adjacent target dividing method based on rainy erosion algorithm |
| CN109034374A (en) * | 2018-07-03 | 2018-12-18 | 苏州中科启慧软件技术有限公司 | The relative depth sequence estimation method of convolutional network is intensively connected to using multi-scale |
| CN109145832A (en) * | 2018-08-27 | 2019-01-04 | 大连理工大学 | Polarimetric SAR image semisupervised classification method based on DSFNN Yu non local decision |
| CN109344837A (en) * | 2018-10-22 | 2019-02-15 | 西安电子科技大学 | A kind of SAR image semantic segmentation method based on depth convolutional network and Weakly supervised study |
| 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 |
| CN110443821A (en) * | 2019-07-09 | 2019-11-12 | 成都理工大学 | Water body detection method and water body detection device based on image |
| CN115223050A (en) * | 2022-04-28 | 2022-10-21 | 湖北工程学院 | Polarized satellite image analysis method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104318246A (en) * | 2014-10-20 | 2015-01-28 | 西安电子科技大学 | Depth self-adaption ridgelet network based polarimetric SAR (Synthetic Aperture Radar) image classification |
| CN104408731A (en) * | 2014-12-09 | 2015-03-11 | 西安电子科技大学 | Region graph and statistic similarity coding-based SAR (synthetic aperture radar) image segmentation method |
| CN104408458A (en) * | 2014-12-09 | 2015-03-11 | 西安电子科技大学 | Ray completion region graph and characteristic learning-based SAR (synthetic aperture radar) image segmentation method |
-
2015
- 2015-10-19 CN CN201510675676.5A patent/CN105374033B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104318246A (en) * | 2014-10-20 | 2015-01-28 | 西安电子科技大学 | Depth self-adaption ridgelet network based polarimetric SAR (Synthetic Aperture Radar) image classification |
| CN104408731A (en) * | 2014-12-09 | 2015-03-11 | 西安电子科技大学 | Region graph and statistic similarity coding-based SAR (synthetic aperture radar) image segmentation method |
| CN104408458A (en) * | 2014-12-09 | 2015-03-11 | 西安电子科技大学 | Ray completion region graph and characteristic learning-based SAR (synthetic aperture radar) image segmentation method |
Non-Patent Citations (2)
| Title |
|---|
| FERNAND MEYER: "Topographic distance and watershed lines", 《SIGNAL PROCESSING》 * |
| 王云艳等: "基于多层反卷积网络的SAR图像分类", 《武汉大学学报·信息科学版》 * |
Cited By (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106529570A (en) * | 2016-10-14 | 2017-03-22 | 西安电子科技大学 | Image classification method based on deep ridgelet neural network |
| CN106529570B (en) * | 2016-10-14 | 2019-06-18 | 西安电子科技大学 | Image classification method based on depth ridge ripple neural network |
| CN106611423A (en) * | 2016-12-30 | 2017-05-03 | 西安电子科技大学 | SAR image segmentation method based on ridge wave filter and deconvolution structural model |
| CN106683102A (en) * | 2016-12-30 | 2017-05-17 | 西安电子科技大学 | SAR image segmentation method based on ridgelet filters and convolution structure model |
| CN106846322A (en) * | 2016-12-30 | 2017-06-13 | 西安电子科技大学 | Based on the SAR image segmentation method that curve wave filter and convolutional coding structure learn |
| CN106683102B (en) * | 2016-12-30 | 2019-07-23 | 西安电子科技大学 | SAR image segmentation method based on ridge ripple filter and convolutional coding structure learning model |
| CN106611423B (en) * | 2016-12-30 | 2019-05-21 | 西安电子科技大学 | SAR image segmentation method based on ridge ripple filter and deconvolution structural model |
| CN106952269A (en) * | 2017-02-24 | 2017-07-14 | 北京航空航天大学 | The reversible video foreground object sequence detection dividing method of neighbour and system |
| CN107341813A (en) * | 2017-06-15 | 2017-11-10 | 西安电子科技大学 | SAR image segmentation method based on structure learning and sketch characteristic inference network |
| CN107292268A (en) * | 2017-06-23 | 2017-10-24 | 西安电子科技大学 | The SAR image semantic segmentation method of quick ridge ripple deconvolution Structure learning model |
| CN108334851A (en) * | 2018-02-12 | 2018-07-27 | 西安电子科技大学 | Based on each to heterogeneous rapid polarization SAR image segmentation method |
| CN108334851B (en) * | 2018-02-12 | 2021-08-03 | 西安电子科技大学 | Rapid polarization SAR image segmentation method based on anisotropic property |
| CN108445466A (en) * | 2018-05-04 | 2018-08-24 | 西安电子科技大学 | A kind of radar adjacent target dividing method based on rainy erosion algorithm |
| CN108445466B (en) * | 2018-05-04 | 2021-09-28 | 西安电子科技大学 | Radar adjacent target segmentation method based on rain erosion algorithm |
| CN109034374A (en) * | 2018-07-03 | 2018-12-18 | 苏州中科启慧软件技术有限公司 | The relative depth sequence estimation method of convolutional network is intensively connected to using multi-scale |
| CN109145832A (en) * | 2018-08-27 | 2019-01-04 | 大连理工大学 | Polarimetric SAR image semisupervised classification method based on DSFNN Yu non local decision |
| CN109344837A (en) * | 2018-10-22 | 2019-02-15 | 西安电子科技大学 | A kind of SAR image semantic segmentation method based on depth convolutional network and Weakly supervised study |
| CN109344837B (en) * | 2018-10-22 | 2022-03-04 | 西安电子科技大学 | SAR image semantic segmentation method based on deep convolutional network and weak supervised learning |
| 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 |
| CN110443821A (en) * | 2019-07-09 | 2019-11-12 | 成都理工大学 | Water body detection method and water body detection device based on image |
| CN110443821B (en) * | 2019-07-09 | 2023-05-05 | 成都理工大学 | Image-based water body detection method and device |
| CN115223050A (en) * | 2022-04-28 | 2022-10-21 | 湖北工程学院 | Polarized satellite image analysis method |
| CN115223050B (en) * | 2022-04-28 | 2023-08-18 | 湖北工程学院 | Polarized satellite image analysis method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105374033B (en) | 2017-11-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105374033A (en) | SAR image segmentation method based on ridgelet deconvolution network and sparse classification | |
| CN103049763B (en) | Context-constraint-based target identification method | |
| CN106228125B (en) | Method for detecting lane lines based on integrated study cascade classifier | |
| CN105427313A (en) | Deconvolutional network and adaptive inference network based SAR image segmentation method | |
| CN102496034B (en) | High-spatial resolution remote-sensing image bag-of-word classification method based on linear words | |
| CN105825502B (en) | A kind of Weakly supervised method for analyzing image of the dictionary study based on conspicuousness guidance | |
| CN106611423B (en) | SAR image segmentation method based on ridge ripple filter and deconvolution structural model | |
| CN104240256A (en) | Image salient detecting method based on layering sparse modeling | |
| CN103729842A (en) | Fabric defect detection method based on local statistical characteristics and overall significance analysis | |
| CN104077613A (en) | Crowd density estimation method based on cascaded multilevel convolution neural network | |
| CN103208001A (en) | Remote sensing image processing method combined with shape self-adaption neighborhood and texture feature extraction | |
| CN104915676A (en) | Deep-level feature learning and watershed-based synthetic aperture radar (SAR) image classification method | |
| CN109766936A (en) | Image change detection method based on information transmitting and attention mechanism | |
| CN103218831A (en) | Video moving target classification and identification method based on outline constraint | |
| CN103034863A (en) | Remote-sensing image road acquisition method combined with kernel Fisher and multi-scale extraction | |
| CN106257498A (en) | Zinc flotation work condition state division methods based on isomery textural characteristics | |
| CN106611420A (en) | SAR image segmentation method based on deconvolution network and sketch direction constraint | |
| CN103020614B (en) | Based on the human motion identification method that space-time interest points detects | |
| CN105608692A (en) | PolSAR image segmentation method based on deconvolution network and sparse classification | |
| CN105335975A (en) | Polarized SAR image segmentation method based on low rank decomposition and histogram statistics | |
| CN106611421A (en) | SAR image segmentation method based on feature learning and sketch line constraint | |
| CN105223561B (en) | Radar ground target discriminator design method based on spatial distribution | |
| CN104573685A (en) | Natural scene text detecting method based on extraction of linear structures | |
| CN106683102A (en) | SAR image segmentation method based on ridgelet filters and convolution structure model | |
| CN104598885A (en) | Method for detecting and locating text sign in street view image |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20171121 |