CN105741286B - The SAR image moving-target shadow extraction method being combined based on width - Google Patents
The SAR image moving-target shadow extraction method being combined based on width Download PDFInfo
- Publication number
- CN105741286B CN105741286B CN201610065517.8A CN201610065517A CN105741286B CN 105741286 B CN105741286 B CN 105741286B CN 201610065517 A CN201610065517 A CN 201610065517A CN 105741286 B CN105741286 B CN 105741286B
- Authority
- CN
- China
- Prior art keywords
- sar image
- target
- moving
- region
- image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Abstract
The invention discloses a kind of SAR image moving-target shadow extraction method being combined based on width, thinking is：Interference processing is carried out to the SAR image data of two adjacency channels after channel-equalization and image registration, obtains interference map of magnitudes and interferometric phase image；Set segmentation threshold, the shadow region extraction in map of magnitudes less than segmentation threshold will be interfered to come out, obtain interference amplitude division figure, and it is K enclosed region to cluster the interference amplitude binary segmentation figure, then the interferometric phase variance of each enclosed region is counted, and interferometric phase variance thresholding is set, then the Closed regions extraction for being higher than the interferometric phase variance thresholding in K enclosed region is come out, obtains the SAR image moving-target shadow region being combined based on width；According to the region for meeting SAR image moving-target shade size range in the SAR image moving-target shadow region being combined based on width described in moving-target experience selection of dimension, as final SAR image moving-target shade.
Description
Technical field
It is the invention belongs to motion platform Radar Moving Target detection technique field, more particularly to a kind of to be combined based on width
SAR image moving-target shadow extraction method, suitable for high-resolution in the case of remotely monitor it is airborne/satellite-borne SAR image at a slow speed
The extraction of moving target shadow region.
Background technology
With the raising of SAR image resolution ratio so that under radar remotely monitor pattern, i.e., low grazing angle is more than incidence angle
In the case of, the shade caused by occlusion effect by target in SAR image is generally existing；In SAR image interpretation, profit
It can be assisted completing target detection and the estimation of target geometric parameter with the shade of target in SAR image.Since moving-target has spy
Fixed radial velocity causes moving-target to deviate actual position in SAR image, and leaves moving-target itself near actual position
Just penetrate shade and the shade of projection, therefore be accurately partitioned into from SAR image moving-target shade and can preferably assist dynamic mesh
Mark detection.
The ground scatter coefficient of dash area is relatively low in SAR image, and echo-signal is close to noise level, and amplitude is relative to week
It is smaller to enclose clutter region, therefore dark areas is presented in dash area in SAR image, and interferometric phase is disorderly and unsystematic, approximation is recognized
It is uniformly distributed to obey.Currently, the extracting method of many dash areas is only built upon amplitude information or SAR in SAR image
Segmentation on gradation of image value information, such as histogram divion method and markov dividing method are all based on SAR image picture
The segmentation of plain gray value is easy by many non-hatched areas, i.e., amplitude or gray scale to be split thresholding close to amplitude or gray scale
And interferometric phase is considered dash area close to zero lower hybrid wave region.
Han Ping et al. is in document《A kind of improved SAR targets and shadow image dividing method based on SVM》(Systems
Engineering and Electronics, 2010,32 (8)) in using SAR image gray value as segmentation feature, first to the back of the body
Scape, target, shade carry out initial segmentation respectively, then support vector machines (SVM) is sent into the output after initial segmentation, and by right
Original SAR image divides iterative cycles Training Support Vector Machines (SVM) again, ultimately forms grader, and then to original SAR image
It is partitioned into background, moving-target, shadow region, this method is only built upon the Shadow segmentation in grey scale pixel value meaning, and being easy will
Compared with lower hybrid wave region segmentation at shadow region.
Suo Zhi is brave et al. in document《Interference SAR shadow extraction and phase compensating method》(Journal od Data
Acquisition & Processing, 2009,24 (3)) in by define spurious correlation coefficient be sharpened the contour correlation of clutter
The boundary in the low correlation zones domain such as region and shade can preferably detect the boundary of shadow region；But since SAR radars are logical
Shadow image between road is not exclusively overlapped so that two parts (including overlapping region) spurious correlation coefficient after being concerned with into row of channels is all
It is very low, it is all divided using this method after increase the error of shade size and life shadow size.
Therefore, more accurate currently without proposing for how preferably to extract the shade of moving-target in SAR image
And calculate easy shadow extraction method.
Invention content
For above the shortcomings of the prior art, it is an object of the invention to propose a kind of SAR being combined based on width
Image motive target shadow extraction method, this kind can solve two based on the SAR image moving-target shadow extraction method that width is combined
A problem, first problem be when relying only on shade amplitude hard thresholding extraction moving-target it is caused certain will be higher than noise level and
Lower hybrid wave region less than amplitude threshold is classified as shadow region, and Second Problem is to only rely on interferometric phase or traditional related coefficient
The Shadow edge loss brought when moving-target shadow region in extraction SAR image, and then can be more accurate using the method for the present invention
Really extract the moving-target shadow region in SAR image.
The present invention main thought be：First, to the SAR of two adjacency channels after channel-equalization and image registration
Image data carries out interference processing, obtains width interference map of magnitudes and a width interferometric phase image；Then, map of magnitudes is interfered in setting
Segmentation threshold is split interference map of magnitudes, and the shadow region in map of magnitudes less than the segmentation threshold of interference map of magnitudes will be interfered to carry
It takes out, obtains a width and interfere amplitude division figure, and it includes K enclosed region to set the interference amplitude binary segmentation figure, so
The interferometric phase variance of each enclosed region is counted afterwards, and sets interferometric phase variance thresholding, then according to the interference of setting
The Closed regions extraction for being higher than the interferometric phase variance thresholding in K enclosed region is come out, obtains base by phase variance thresholding
In the SAR image moving-target shadow region that width is combined；Finally, it is combined based on width according to described in moving-target experience selection of dimension
SAR image moving-target shadow region in meet the region of SAR image moving-target shade size range, scheme as final SAR
As moving-target shadow region.
To reach above-mentioned technical purpose, the present invention is realised by adopting the following technical scheme.
A kind of SAR image moving-target shadow extraction method being combined based on width, is included the following steps：
Step 1, SAR image is obtained, chooses any two adjacency channel in the SAR image, and to described two adjacent
The corresponding SAR image data in channel carry out channel-equalization respectively, obtain the SAR image equalization data and second of first passage
The SAR image equalization data in channel, then using the SAR image equalization data of first passage as refer to channel SAR image data,
Image registration is carried out to the SAR image equalization data of second channel, obtains the SAR image registration data of second channel；
Step 2, interference processing is carried out to reference channel SAR image data and the SAR image registration data, obtained respectively
Interfere map of magnitudes and interferometric phase image；
Step 3, the segmentation threshold of interference map of magnitudes is obtained, and interference map of magnitudes is converted into gray level image, is then extracted
It is less than the shadow region of the segmentation threshold in gray level image, then using the corresponding shadow region extracted as interference amplitude two
It is worth segmentation figure；
Step 4, interferometric phase variance thresholding is set, the picture for being more than the segmentation threshold in amplitude binary segmentation figure will be interfered
Vegetarian refreshments cluster is K enclosed region, and the pixel that the K enclosed region respectively contains corresponds to the spy in interferometric phase image respectively
The numerical value of fixation vegetarian refreshments, the pixel that wherein each enclosed region includes respectively equal to corresponds to specific pixel in interferometric phase image
Point numerical value, and using the numerical value of specific pixel point in interferometric phase image as correspondence enclosed region in pixel interferometric phase,
Then the corresponding pixel interferometric phase variance of each enclosed region is counted, further according to the interferometric phase variance thresholding of setting,
The Closed regions extraction for being higher than the interferometric phase variance thresholding in K enclosed region is come out, will be extracted higher than described
The enclosed region of interferometric phase variance thresholding, as the SAR image moving-target shadow region being combined based on width；
Step 5, SAR image moving-target shade size range is obtained, the SAR image being combined based on width is then chosen and moves mesh
The shadow region for meeting SAR image moving-target shade size range in mark shadow region, it is cloudy as final SAR image moving-target
Shadow.
Compared with prior art, the present invention haing the following advantages：
First, the present invention is by that by the interference amplitude in two channels and interferometric phase use in conjunction, can improve and rely only on the moon
Shadow amplitude hard thresholding is caused when extracting moving-target to return by certain higher than noise level and less than the lower hybrid wave regions of amplitude threshold
For shadow region problem, can also solve to only rely on moving-target shade in interferometric phase or traditional related coefficient extraction SAR image
The Shadow edge loss problem brought when region, and then can more accurately be extracted in SAR image using the method for the present invention
Moving-target shadow region.
Second, the method for the present invention can preferably retain the shape information and dimension information of target shadow, can be follow-up dynamic
Moving-target and its shade correspondence in target detection, and the estimation of corresponding moving-target geometric dimension provide more accurate item
Part.
Description of the drawings
The present invention is described in further details with reference to the accompanying drawings and detailed description.
Fig. 1 is a kind of SAR image moving-target shadow extraction method flow diagram being combined based on width of the present invention；
Fig. 2 is the SAR scene graph for shadow extraction of the present invention；
Fig. 3 is the present invention merely with interference amplitude threshold shadow extraction design sketch；
Fig. 4 is the scene interferometric phase image of the present invention；
Fig. 5 is the shadow extraction design sketch of the amplitude and phase combining of the present invention；
Fig. 6 is the shadow extraction result figure of the present invention rejected using target experience size after non-hatched area.
Specific implementation mode
Referring to Fig.1, it is a kind of SAR image moving-target shadow extraction method flow diagram being combined based on width of the present invention；
A kind of SAR image moving-target shadow extraction method being combined based on width of the present invention, is included the following steps：
Step 1, SAR image is obtained, chooses any two adjacency channel in the SAR image, and to described two adjacent
The corresponding SAR image data in channel carry out channel-equalization respectively, obtain the SAR image equalization data and second of first passage
The SAR image equalization data in channel, then using the SAR image equalization data of first passage as refer to channel SAR image data,
Image registration is carried out to the SAR image equalization data of second channel, obtains the SAR image registration data of second channel.
Specifically, SAR image is obtained, chooses any two adjacency channel in the SAR image, and to described two adjacent
The corresponding SAR image data in channel carry out channel-equalization respectively, obtain the SAR image equalization data and second of first passage
The SAR image equalization data in channel, then using the SAR image equalization data of first passage as refer to channel SAR image data,
Image registration is carried out to the SAR image equalization data of second channel, the present embodiment carries out the pixel of interchannel using cross-correlation method
Grade registration, uniformly cuts into Nu tile data successively by the SAR image equalization data of second channel first, arbitrary to choose wherein
The gray scale value matrix of one tile data and reference channel SAR image data does two-dimensional cross correlation processing, obtains second channel
Selected tile data corresponding pixel gray level value matrix T (x, y) and reference at position (x, y) in SAR image equalization data
Gray value similarity degree relational expression C (u, v) of each pixel at offset (u, v) in the SAR image data of channel：
Wherein, x indicates that orientation pixel ordinal number in SAR image, y indicate SAR image middle-range descriscent pixel ordinal number,
(x, y) indicates a pixel position in SAR image, and C (u, v) indicates selected in the SAR image equalization data of second channel
Tile data at position (x, y) corresponding pixel gray level value matrix T (x, y) with it is each in reference channel SAR image data
Gray value similarity degree relational expression of a pixel at offset (u, v), u indicate that orientation offset pixels point number, v indicate
Distance deviates number to pixel, and the value range of u is：0≤u≤Na, Na indicate SAR image upper position to pixel
Number, the value range of u are：0≤u≤Nr；Nr indicate distance in SAR image to pixel number, T (x, y) indicates that second is logical
Selected tile data corresponding pixel gray level value matrix at position (x, y), f (x, y) table in the SAR image equalization data in road
Show pixel gray level value matrix of the reference channel SAR image data at position (x, y).
When selected tile data corresponding pixel ash at position (x, y) in the SAR image equalization data of second channel
The pixel gray level value matrix f (x, y) of angle value matrix T (x, y) and reference channel SAR image data at position (x, y) is
In the SAR image equalization data in two channels selected tile data at position (x, y) corresponding pixel gray level value matrix T (x,
Y) displacement is (x0,y0) at when matching, C (u, v) can be in C (x0,y0) at there is correlation peak, and by two dimension mutually
Corresponding displacement (the x of correlation peak obtained after the processing of pass0,y0) it is used as thick displacement vector (x0,y0)；Then referring to described thick
Displacement vector (x0,y0) acquisition process, by N in the SAR image equalization data of second channeluA tile data is corresponding
Gray scale value matrix obtains N after doing two-dimensional cross correlation processing with the gray scale value matrix of reference channel SAR image data respectivelyuA phase
Peak coordinate shift amount is closed, and then obtains the SAR image registration data of second channel；The NuA relevant peaks coordinate shift amount is
(x01,y01)、(x02,y02)、…、(x0a,y0a)、…、a∈{1,2,…,Nu}；Then to the Nu
After a relevant peaks coordinate shift amount carries out statistical average and rounding calculating, the displacement vector estimated value for registration is obtainedAccording to the displacement vector estimated value for registrationEntirety is done to the SAR image equalization data of second channel
Translation, can improve the signal coherency between two adjacency channels.
Then, the SAR image registration data of second channel and reference channel SAR image data are transformed into distance-respectively
The reference of Doppler domain, the SAR image registration data and distance-Doppler domain that obtain the second channel in distance-Doppler domain is logical
Road SAR image data, and the ginseng of the SAR image registration data and distance-Doppler domain of the second channel in distance-Doppler domain
The corresponding each distance-Doppler unit of channel SAR image data is examined, it is a pair of with the pixel one in SAR image respectively
It answers, while the corresponding distance-Doppler unit composition observation data vector of the pixel in SAR image, finally uses adaptive
Ro-vibrational population compensation method realizes channel-equalization to the observation data vector.
Step 2, interference processing is carried out to reference channel SAR image data and the SAR image registration data, obtained respectively
Interfere map of magnitudes and interferometric phase image.
Specifically, it is s that reference channel SAR image data, which are set separately,1, the SAR image registration data of second channel is s2,
Then to reference channel SAR image data s1With the SAR image registration data s of second channel2Interference processing is carried out, is obtained respectively
Interfere map of magnitudes SamWith interferometric phase image San, expression formula is respectively：
Wherein, abs () expressions take magnitude operations, angle () expressions to take phase operation, ()*Indicate conjugate operation,
s1Indicate reference channel SAR image data, s2Indicate the SAR image registration data of second channel.
Step 3, the segmentation threshold of interference map of magnitudes is obtained, and interference map of magnitudes is converted into gray level image, is then used
It is less than the shadow region of the segmentation threshold in amplitude division method extraction gray level image in image procossing, then correspondence is extracted
The shadow region come is as interference amplitude binary segmentation figure.
Specifically, the segmentation threshold T of interference map of magnitudes is obtained using inter-class variance maximum principleam, then map of magnitudes will be interfered
SamIt is converted into gray level image Gam, then by nTamAs segmentation gray level image GamAdjustable threshold, and use image procossing
In amplitude division method extraction gray level image in be less than the shadow region of the segmentation threshold, the present embodiment is calculated using single order Ostu
Method or second order Ostu algorithms are to gray level image GamMiddle gray value is less than nTamShadow region extract, then will extract
Correspondence shadow region, as interference amplitude binary segmentation image image1, i.e.,：
Wherein, gamIndicate gray level image GamIn any one pixel gray value；Image1 is and gray level image Gam
The identical two values matrix of dimension, and with gray level image GamIn pixel correspond；n·TamIndicate segmentation gray level image Gam
Adjustable threshold, n>0, n indicates the segmentation threshold adjustment factor of setting, TamIt indicates to utilize the acquisition of inter-class variance maximum principle
Segmentation threshold.
Opening operation in morphologic filtering processing method and closed operation are carried out to interference amplitude binary segmentation image image1
It handles, the small pixel group in interference amplitude binary segmentation image image1 and smooth shadow region profile after elimination Shadow segmentation,
Also preferably retain the marginal information in dashed horizontal region simultaneously.
Step 4, interferometric phase variance thresholding is set, the picture for being more than the segmentation threshold in amplitude binary segmentation figure will be interfered
Vegetarian refreshments cluster is K enclosed region, and the pixel that the K enclosed region respectively contains corresponds to the spy in interferometric phase image respectively
Fixation vegetarian refreshments, the pixel that wherein each enclosed region includes correspond to the numerical value of specific pixel point in interferometric phase image respectively,
And using the numerical value of specific pixel point in interferometric phase image as the pixel interferometric phase in corresponding enclosed region, then statistics is every
The corresponding pixel interferometric phase variance of one enclosed region, further according to the interferometric phase variance thresholding of setting, by K closed area
Closed regions extraction in domain higher than the interferometric phase variance thresholding comes out, and is higher than the interferometric phase variance by what is extracted
The enclosed region of thresholding, as the SAR image moving-target shadow region being combined based on width.
Specifically, interferometric phase variance thresholding is set, is more than the segmentation threshold in amplitude binary segmentation figure by interfering
Pixel cluster is K enclosed region, and the pixel that the K enclosed region respectively contains corresponds in interferometric phase image respectively
The numerical value of specific pixel point, the pixel that wherein each enclosed region includes respectively equal to corresponds to specific picture in interferometric phase image
The numerical value of vegetarian refreshments, and interfere phase using the numerical value of specific pixel point in interferometric phase image as the pixel in corresponding enclosed region
Position, then counts the corresponding pixel interferometric phase variance of each enclosed region, further according to the interferometric phase variance door of setting
Limit comes out the Closed regions extraction for being higher than the interferometric phase variance thresholding in K enclosed region, the K enclosed region
The pixel number respectively contained is followed successively by：m1、m2、…、mk、…、mM, k ∈ { 1,2 ..., K }, k-th of enclosed region include
mkA pixel is denoted as：i∈{1,2,…,mk, pkiIt indicates in k-th of enclosed region
Ith pixel point, and then the pixel interferometric phase variance V of k-th of enclosed region is calculatedk, expression formula is：
Vk=var (San(pki))
Wherein, pkiIndicate the ith pixel point in k-th of enclosed region, i ∈ { 1,2 ..., mk, mkIt indicates to close for k-th
The pixel number that region includes is closed, var (*) indicates variance operator, San(*) indicates interferometric phase.
Since the interferometric phase in clutter region in SAR image is close to zero, variance is also smaller, and shadow region and close to noise
The interferometric phase obedience of horizontal zone is uniformly distributed [- π π], and variance is closeTherefore the interference to the K enclosed region
Phase variance sets an interferometric phase variance thresholdingFurther according to the interference of setting
Phase variance thresholding Tvar, the interferometric phase variance thresholding T will be higher than in K enclosed regionvarClosed regions extraction come out,
The enclosed region higher than the interferometric phase variance thresholding that will be extracted, as the SAR image moving-target being combined based on width
Shadow region image2, expression formula are：
Wherein, TvarIndicate that the interferometric phase variance thresholding of setting, k ∈ { 1,2 ..., K }, K indicate interference amplitude two-value point
Cut the enclosed region number that figure includes, VkIndicate the pixel interferometric phase variance of k enclosed region；Image2 indicates to be based on width
The SAR image moving-target shadow region being combined, and be two-value identical with interference amplitude binary segmentation image image1 dimensions
Matrix.
Step 5, SAR image moving-target shade size range is obtained according to the experience size of moving-target, then chooses and is based on
The shadow region for meeting SAR image moving-target shade size range in the SAR image moving-target shadow region that width is combined, as
Final SAR image moving-target shade.
Specifically, according to the experience size of ground moving target in reality, SAR image moving-target shade size range is obtained,
Remember that the length of moving-target in SAR image is respectively a=[Tami,Tama], b=[Tbmi,Tbma], h=[Thmi,Thma], then
Choose the shade for meeting SAR image moving-target shade size range in the SAR image moving-target shadow region being combined based on width
Region, as final SAR image moving-target shade, selection condition is：
Sl∈[min(Tbmi,Tami),max(Tbma,Tama)]
Sw∈[min(Tbmi,Tami)+Thmi·tan(α)·sin(α),max(Tbma,Tama)+Thma·tan(α)·sin
(α)]
Wherein, SlIndicate the orientation width of SAR image moving-target shade, SwIndicate the distance of SAR image moving-target shade
To width, α indicates that corresponding SAR radar platforms downwards angle of visibility when SAR radar imageries, max (*) expressions are maximized, min (*) table
Show and be minimized, sin (*) indicates mathematically to seek the operator of triangle sine value, and tan (*) expressions are mathematically seeking triangle just
The operator cut, TamiIndicate the minimum length value of moving-target in SAR image, TamaMoving-target most greatly enhances in expression SAR image
Angle value, TbmiIndicate the minimum width value of moving-target in SAR image, TbmaIndicate the maximum width value of moving-target in SAR image,
ThmiIndicate the minimum height values of moving-target in SAR image, ThmaIndicate the maximum width value of moving-target in SAR image.
In the shadow regions SAR being combined based on width obtained from step 4, selection meets SAR image moving-target shade ruler
The region of very little range, and then final SAR image moving-target shade is obtained, subsequently to continue to confirm moving-target and corresponding shade
Relationship provides precondition.
Effect of the present invention can be further illustrated by following emulation experiment.
(1) experimental situation
Various simulation parameters used in the present invention are as follows：
It is reference channel, interchannel with channel 1 that triple channel SAR radar moving targets systems, which are emulated, using multicast pattern
Baseline length is 1m, and carrier frequency 10GHz, pulse recurrence frequency 400Hz, SAR radar platform speed is 200m/s, SAR radars
Corresponding SAR radar platforms downwards angle of visibility is 76 ° when imaging；The minimum oblique distance of simulating scenes to SAR radar platforms is 20810m, field
Scape moving-target is 1m in orientation and apart from upward resolution ratio, and miscellaneous noise ratio (Clutter-to- is inputted before two-dimentional pulse pressure
Noise Rate, CNR) CNR=-35dB, four moving-target shadow regions are simulated in simulating scenes.
(2) experiment content and result
Specific scene is emulated using range Doppler (RD) imaging algorithm, imaging results are as shown in Fig. 2, Fig. 2
For the SAR scene graph for shadow extraction of the present invention；As can be seen from Figure 2 four moving-target shadow regions, each rectangle
The shadow region that closely stationary object occurs in frame, and large stretch of river and low energy clutter region.
Interference amplitude binary segmentation figure such as Fig. 3 after given interference amplitude threshold indicates that Fig. 3 is the present invention merely with dry
Relate to amplitude threshold shadow extraction design sketch；The interferometric phase in the dashed horizontal region in SAR image is as shown in figure 4, in Fig. 4 simultaneously
Also include the interferometric phase of four targets, shadow region interferometric phase is disorderly and unsystematic as can be seen from Figure 4, and phase is interfered in clutter region
Position is almost 0, and since interference amplitude threshold belongs to hard segmentation, interferometric phase is not mixed and disorderly enough close to 0 or interferometric phase in Fig. 4
Low energy clutter region be extracted.
The present invention, close to the principle of noise phase variance, gives extraction of the present invention using shadow region interferometric phase variance
The results are shown in Figure 5 for shade, and Fig. 5 is the shadow extraction design sketch of the amplitude and phase combining of the present invention；Wherein, comparison diagram 3,
Fig. 5 deletes a part of non-hatched area according to interferometric phase variance threshold values, it can be seen that representated by the white area in Fig. 5
Shade of white region in the ratio Fig. 3 of shadow region is few, meanwhile, in figure 5 it can be seen that the shadow region irised out in Fig. 2 is had
Effect retains, and marginal information and shape information extraction are intact.Fig. 6 is by the calculated correspondence of target experience size in step 5
It is after the screening of shade size range as a result, Fig. 6 be the present invention utilize target experience size reject non-hatched area after the moon
Shadow extracts result figure, can see that equidimension larger non-targeted shadow region in river is rejected from Fig. 6, remaining area is all dynamic
The corresponding possible shadow region of target.
In conclusion the method for the present invention under long-range SAR radar operation modes merges on the basis of interfering amplitude extraction
The characteristics of shadow region interferometric phase variance is close to noise phase variance eliminates the non-of the hard thresholding extraction shadow band of amplitude
Shadow region, and it has been effectively kept the shape information of shadow region, experiment simulation also demonstrates the method for the present invention and is remotely supervising
It is more accurate to the shadow region extraction of moving-target in High Resolution SAR image under optionally.
Obviously, various changes and modifications can be made to the invention without departing from essence of the invention by those skilled in the art
God and range.In this way, if these modifications and changes of the present invention belongs to the range of the claims in the present invention and its equivalent technologies
Within, then the present invention is also intended to include these modifications and variations.
Claims (6)
1. a kind of SAR image moving-target shadow extraction method being combined based on width, which is characterized in that include the following steps：
Step 1, SAR image is obtained, chooses any two adjacency channel in the SAR image, and to described two adjacency channels
Corresponding SAR image data carry out channel-equalization respectively, obtain the SAR image equalization data and second channel of first passage
SAR image equalization data, then using the SAR image equalization data of first passage as channel SAR image data are referred to, to the
The SAR image equalization data in two channels carries out image registration, obtains the SAR image registration data of second channel；
Step 2, interference processing is carried out to reference channel SAR image data and the SAR image registration data, obtains interference respectively
Map of magnitudes and interferometric phase image；
Step 3, the segmentation threshold of interference map of magnitudes is obtained, and interference map of magnitudes is converted into gray level image, then extracts gray scale
It is less than the shadow region of the segmentation threshold in image, then using the corresponding shadow region extracted as interference amplitude two-value point
Cut figure；
Step 4, interferometric phase variance thresholding is set, the pixel for being more than the segmentation threshold in amplitude binary segmentation figure will be interfered
Cluster is K enclosed region, and the pixel that the K enclosed region respectively contains corresponds to the specific picture in interferometric phase image respectively
The numerical value of vegetarian refreshments, the pixel that wherein each enclosed region includes is respectively equal to specific pixel point in corresponding interferometric phase image
Numerical value, and using the numerical value of specific pixel point in interferometric phase image as the pixel interferometric phase in corresponding enclosed region, then
The corresponding pixel interferometric phase variance of each enclosed region is counted, further according to the interferometric phase variance thresholding of setting, by K
Closed regions extraction in enclosed region higher than the interferometric phase variance thresholding comes out, and is higher than the interference phase by what is extracted
The enclosed region of position variance thresholding, as the SAR image moving-target shadow region being combined based on width；
Step 5, SAR image moving-target shade size range is obtained according to the experience size of moving-target, then chooses and is based on width phase
The region for meeting SAR image moving-target shade size range in united SAR image moving-target shadow region, as final
SAR image moving-target shadow region.
2. a kind of SAR image moving-target shadow extraction method being combined based on width as described in claim 1, feature are existed
In in step 2, the interference map of magnitudes and interferometric phase image further include：
Interference map of magnitudes is denoted as Sam, interferometric phase image is denoted as San, expression formula is respectively：
Wherein, abs () expressions take magnitude operations, angle () expressions to take phase operation, ()*Indicate conjugate operation, s1Table
Show reference channel SAR image data, s2Indicate the SAR image registration data of second channel.
3. a kind of SAR image moving-target shadow extraction method being combined based on width as described in claim 1, feature are existed
In, it is in step 3, described to extract the shadow region for being less than the segmentation threshold in gray level image, further include：
The segmentation threshold T of interference map of magnitudes is obtained using inter-class variance maximum principleam, then will interference map of magnitudes SamIt is converted into gray scale
Image Gam, then by nTamAs segmentation gray level image GamAdjustable threshold, and use single order Ostu algorithms or second order
Ostu algorithms are to gray level image GamMiddle gray value is less than nTamShadow region extract；Wherein, nTamIndicate segmentation ash
Spend image GamAdjustable threshold, n > 0, n indicate setting segmentation threshold adjustment factor.
4. a kind of SAR image moving-target shadow extraction method being combined based on width as described in claim 1, feature are existed
In in step 4, the SAR image moving-target shadow region being combined based on width further includes：
The SAR image moving-target shadow region being combined based on width is denoted as image2, expression formula is：
Wherein, k ∈ { 1,2 ..., K }, K indicate the enclosed region number that interference amplitude binary segmentation figure includes, VkIndicate k closure
The pixel interferometric phase variance in region, TvarIndicate the interferometric phase variance threshold values of setting, and
Δ T indicates the adjustable parameter range of interferometric phase variance threshold values.
5. a kind of SAR image moving-target shadow extraction method being combined based on width as claimed in claim 4, feature are existed
In the VkIndicate that the pixel interferometric phase variance of k enclosed region, expression formula are：
Vk=var (San(pki))
Wherein, pkiIndicate the ith pixel point in k-th of enclosed region, i ∈ { 1,2 ..., mk, mkIndicate k-th of enclosed region
Including pixel number, var (*) indicate variance operator, San(*) indicates interferometric phase.
6. a kind of SAR image moving-target shadow extraction method being combined based on width as described in claim 1, feature are existed
In in steps of 5, described choose meets SAR image moving-target the moon in the SAR image moving-target shadow region being combined based on width
The region of shadow size range, selection condition are：
Sl∈[min(Tbmi, Tami), max (Tbma, Tama)]
Sw∈[min(Tbmi, Tami)+ThmiTan (α) sin (α), max (Tbma, Tama)+Thma·tan(α)·sin(α)]
Wherein, SlIndicate the orientation width of SAR image moving-target shade, SwIndicate the distance of SAR image moving-target shade to width
Degree, α indicate that corresponding SAR radar platforms downwards angle of visibility when SAR radar imageries, max (*) expressions are maximized, and min (*) expressions take
Minimum value, sin (*) expressions mathematically ask the operator of triangle sine value, tan (*) expressions mathematically to seek triangle tangent
Operator, TamiIndicate the minimum length value of moving-target in SAR image, TamaIndicate the maximum length value of moving-target in SAR image,
TbmiIndicate the minimum width value of moving-target in SAR image, TbmaIndicate the maximum width value of moving-target in SAR image, ThmiIt indicates
The minimum height values of moving-target in SAR image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610065517.8A CN105741286B (en) | 2016-01-29 | 2016-01-29 | The SAR image moving-target shadow extraction method being combined based on width |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610065517.8A CN105741286B (en) | 2016-01-29 | 2016-01-29 | The SAR image moving-target shadow extraction method being combined based on width |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105741286A CN105741286A (en) | 2016-07-06 |
| CN105741286B true CN105741286B (en) | 2018-11-09 |
Family
ID=56247972
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610065517.8A Active CN105741286B (en) | 2016-01-29 | 2016-01-29 | The SAR image moving-target shadow extraction method being combined based on width |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105741286B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108051772B (en) * | 2017-10-25 | 2021-12-07 | 西南电子技术研究所（中国电子科技集团公司第十研究所） | Amplitude-phase combined method for measuring incoming wave azimuth information |
| CN107883917B (en) * | 2017-11-20 | 2021-02-02 | 中国科学院遥感与数字地球研究所 | High-voltage line tower height extraction method based on multi-temporal shadow difference |
| CN111126314A (en) * | 2019-12-26 | 2020-05-08 | 杭州中科先进技术研究院有限公司 | Passenger flow statistical visual method and system based on 3D point cloud data |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102980533A (en) * | 2009-07-03 | 2013-03-20 | 株式会社高永科技 | Method for inspecting measurement object |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2011034178A (en) * | 2009-07-30 | 2011-02-17 | Sony Corp | Apparatus and method for image processing, and program |
-
2016
- 2016-01-29 CN CN201610065517.8A patent/CN105741286B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102980533A (en) * | 2009-07-03 | 2013-03-20 | 株式会社高永科技 | Method for inspecting measurement object |
Non-Patent Citations (1)
| Title |
|---|
| 基于联合像素干涉统计特性的三通道动目标检测方法;吕孝雷等;《电子学报》;20081231;第36卷(第12期);2319-2323 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105741286A (en) | 2016-07-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104851097B (en) | The multichannel SAR GMTI methods aided in based on target shape and shade | |
| White | Change detection in SAR imagery | |
| CN101650439B (en) | Method for detecting change of remote sensing image based on difference edge and joint probability consistency | |
| CN102778680B (en) | Method for imaging uniformly accelerated motion rigid group targets based on parameterization | |
| CN102402685B (en) | Method for segmenting three Markov field SAR image based on Gabor characteristic | |
| CN108961255B (en) | Sea-land noise scene segmentation method based on phase linearity and power | |
| CN105427314A (en) | Bayesian saliency based SAR image target detection method | |
| CN105741286B (en) | The SAR image moving-target shadow extraction method being combined based on width | |
| CN107274401A (en) | A kind of High Resolution SAR Images Ship Detection of view-based access control model attention mechanism | |
| CN107561533B (en) | A kind of C-band satellite-borne synthetic aperture radar motive target imaging method | |
| CN102903102A (en) | Non-local-based triple Markov random field synthetic aperture radar (SAR) image segmentation method | |
| CN106908781A (en) | The velocity acquisition methods of the linear motion moving-target based on single channel circular track SAR | |
| CN103871039A (en) | Generation method for difference chart in SAR (Synthetic Aperture Radar) image change detection | |
| CN105427301B (en) | Based on DC component than the extra large land clutter Scene Segmentation estimated | |
| CN101546429B (en) | Detection method of small target of feature space | |
| CN105974412A (en) | Target feature extraction method used for synthetic aperture radar | |
| CN102496142B (en) | SAR (synthetic aperture radar) image segmentation method based on fuzzy triple markov fields | |
| Ran et al. | Simultaneous range and cross-range variant phase error estimation and compensation for highly squinted SAR imaging | |
| CN101482969A (en) | SAR image speckle filtering method based on identical particle computation | |
| CN106054187A (en) | High squint SAR curvilinear path wavenumber domain imaging method based on slope distance model | |
| CN104680536A (en) | Method for detecting SAR image change by utilizing improved non-local average algorithm | |
| CN109188436B (en) | Efficient bistatic SAR echo generation method suitable for any platform track | |
| Yin et al. | A method for automatic target recognition using shadow contour of SAR image | |
| CN104680181A (en) | SAR image super-pixel segmentation method based on likelihood ratio features | |
| CN105116410B (en) | The interferometric phase image adaptive filter algorithm matched based on linear model |
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 |