CN103869290B - Based on the phase unwrapping winding method of large-scale data weighting branch tangent line - Google Patents

Based on the phase unwrapping winding method of large-scale data weighting branch tangent line Download PDF

Info

Publication number
CN103869290B
CN103869290B CN201410105328.XA CN201410105328A CN103869290B CN 103869290 B CN103869290 B CN 103869290B CN 201410105328 A CN201410105328 A CN 201410105328A CN 103869290 B CN103869290 B CN 103869290B
Authority
CN
China
Prior art keywords
network
sub
residual
point
residual point
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
Application number
CN201410105328.XA
Other languages
Chinese (zh)
Other versions
CN103869290A (en
Inventor
李真芳
王志斌
刘艳阳
索志勇
李锦伟
杨桃丽
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xidian University
Original Assignee
Xidian University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Xidian University filed Critical Xidian University
Priority to CN201410105328.XA priority Critical patent/CN103869290B/en
Publication of CN103869290A publication Critical patent/CN103869290A/en
Application granted granted Critical
Publication of CN103869290B publication Critical patent/CN103869290B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/88—Radar or analogous systems specially adapted for specific applications
    • G01S13/89—Radar or analogous systems specially adapted for specific applications for mapping or imaging
    • G01S13/90—Radar or analogous systems specially adapted for specific applications for mapping or imaging using synthetic aperture techniques, e.g. synthetic aperture radar [SAR] techniques
    • G01S13/9021—SAR image post-processing techniques
    • G01S13/9023—SAR image post-processing techniques combined with interferometric techniques
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
    • G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section

Landscapes

  • Engineering & Computer Science (AREA)
  • Remote Sensing (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Physics & Mathematics (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • General Physics & Mathematics (AREA)
  • Electromagnetism (AREA)
  • Image Analysis (AREA)

Abstract

The invention discloses a kind of phase unwrapping winding method based on large-scale data weighting branch tangent line, overcome prior art and phase unwrapping is carried out around the slow problem of processing speed to large-scale data, efficiently solve prior art and adopt rectangular partition and the discontinuous problem of phase place that causes.Performing step of the present invention is: 1. input data; 2. residual identification; 3. potential branch tangent line path is set; 4. build sub-network; 5. add ground node; 6. find out uneven sub-network; 7. correct balance sub-network; 8. judge solvability; 9. arrange branch tangent line; 10. phase gradient integration.The present invention has and carries out phase unwrapping processing speed to large-scale data and conciliate soon and twine the high advantage of precision, may be used for phase unwrapping under large-scale data around process.

Description

Based on the phase unwrapping winding method of large-scale data weighting branch tangent line
Technical field
The invention belongs to communication technical field, further relate to the phase unwrapping winding method of the interference synthetic aperture radar (InterferometrySyntheticApertureRadar, InSAR) in the radar exploration technique field based on large-scale data weighting branch tangent line.The present invention can be used for carrying out quick unwrapping process to the interferometric phase image under large-scale data, and solution prior art phase unwrapping under large-scale data is slow around processing speed, the problem of low precision.
Background technology
Phase unwrapping around the committed step being InSAR data processing, phase unwrapping around result directly determine the quality of digital elevation model (DigitalElevationModel, DEM).Along with the development of synthetic-aperture radar (SyntheticApertureRadar, SAR) technology, the data volume that SAR obtains constantly increases, and this becomes large by making the scale of phase unwrapping winding number certificate; And along with the raising of resolution, SAR image shadow effect is more obvious, and residual the counting of interferometric phase also increases, and phase unwrapping is increasing around the difficulty of process thereupon.
Under large-scale data, existing phase unwrapping winding method processing speed is slow, adds InSAR data processing time, and existing method adopts strategy large-scale data being carried out to rectangular partition, causes solution to twine the discontinuous problem of phase place.
The frequency domain algorithm of a kind of many baselines/multiband phase unwrapping that PLA University of Science and Technology for National Defense proposes in its patented claim " many baselines/multiband interferometric phase solution twines frequency domain fast algorithm " (number of patent application: 201110312439, publication number: 102621549A).Time domain phase gradient value is transformed into frequency domain process by this algorithm, calculates the approximate value of the fourier coefficient meeting irrotationality condition.Although this invention is not when affecting precision, decrease the calculated amount of phase unwrapping algorithm, the deficiency still existed is, there is the problem of speed slow and low precision when carrying out unwrapping process to large-scale data.
ZhangK., GeL. with HuZ. at the paper delivered " Phaseunwrappingforverylargeinterferometricdatasets " (" IEEETransactionsonGeoscienceAndRemoteSensing ", 2011,10 (49): 4048-4061) propose in and the algorithm of phase unwrapping around process is carried out to large-scale data.This algorithm utilizes the shortest optimisation strategy of overall branch tangential length to arrange a tangent line, carries out phase unwrapping around process to whole interferometric phase image.The deficiency that this algorithm exists is that the network optimization process in algorithm, in the unbalanced situation of netinit, needs to carry out iterative processing, and this will cause this Algorithms T-cbmplexity to increase, slower to processing speed during large-scale data process; And this algorithm adopts rectangular partition strategy to split network, and the unwrapping phase place between each sub-block exists discontinuous problem.
Summary of the invention
The present invention is directed to large-scale data phase unwrapping that above-mentioned prior art exists around speed the slow and discontinuous problem of unwrapping phase place, propose a kind of phase unwrapping winding method based on large-scale data weighting branch tangent line, the path of potential branch tangent line is set by the quality information of interferometric phase image, then utilizes these potential branch tangent lines to build multiple tangent line sub-networks of view picture interferometric phase image.After detecting the balance of each sub-network and solvability and correct, utilize minimum cost flow algorithm to optimize each sub-network, utilize optimum results to determine the optimum access path of a tangent line, realize the quick unwrapping process of interferometric phase image.
For achieving the above object, key step of the present invention is as follows:
(1) data are inputted:
(1a) the interferometric phase diagram data of input after interference treatment;
(1b) quality information data of interferometric phase image is inputted.
(2) residual point is identified:
(2a) by first pixel I in interferometric phase image 1,1position, as the position of extracting pixel-phase value in interferometric phase image; Wherein, I 1,1represent that in interferometric phase image, line number and row number are the pixel of 1;
(2b) four pixel I in interferometric phase image are extracted successively n,Mi n, M+1i n+1, Mi n+1, M+1phase value, will pixel I be extracted n+1, Mposition mark be the position of extracting pixel-phase value next time, the phase value of four pixels of extraction generates an arest neighbors closed-loop path; Wherein, I n,Mrepresent the pixel that in interferometric phase image, line number is N, row number are M, I n, M+1represent the pixel that in interferometric phase image, line number is N, row number are M+1, I n+1, Mrepresent the pixel that in interferometric phase image, line number is N+1, row number are M, I n+1, M+1represent the pixel that in interferometric phase image, line number is N+1, row number are M+1;
(2c) along clockwise direction integration is carried out to arest neighbors closed-loop path, if integral result is 2 π, by the pixel I of four in closed-loop path n,Mi n, M+1i n+1, Mi n+1, M+1center in the interferometric phase image of place is labeled as just residual point; If integral result is-2 π, by the pixel I of four in closed-loop path n,Mi n, M+1i n+1, Mi n+1, M+1center in the interferometric phase image of place is labeled as negative residual point;
(2d) the pixel I extracted is judged n+1, M+1be whether last pixel of interferometric phase image, if so, perform step (3); Otherwise, perform step (2b).
(3) potential branch tangent line path is set:
(3a) integer range is set as by 30 ~ 150;
(3b) for the residual point that density is large, a less integer is chosen in integer range, for the residual point that density is little, in integer range, choose a larger integer, using the integer chosen as the maximum distance between just residual point and negative residual point and with just residual be connected maximumly bear residual counting;
(3c) position of the residual point marked according to step (2c), around just residual at first, the from the close-by examples to those far off border of the negative residual point of search and interferometric phase image; Judge whether the border three of just residual point, negative residual point and interferometric phase image meets to impose a condition, if met, perform step (3e); Otherwise, perform step (3h);
(3d) around the just residual point of the next one, the from the close-by examples to those far off border of the negative residual point of search and interferometric phase image; Judge whether the border three of just residual point, negative residual point and interferometric phase image meets to impose a condition, if met, perform step (3e); Otherwise, perform step (3h);
(3e) by the residual point that just residual point and all negative residual composition corresponding with it match, network to be optimized is formed by the residual of all pairings;
(3f) weighted optimization network to be optimized, chooses optimum access path between residual point;
(3g) in interferometric phase image, all nodes of optimal path process are connected, using connecting line as potential branch tangent line path;
(3h) judge whether just residual point is last just residual point, if so, perform step (4); Otherwise, perform step (3d).
(4) sub-network is built:
(4a) according to the sequence of positions of just residual point in interferometric phase image, first just residual point and the just residual all negative residual point be connected therewith is chosen;
(4b) a newly-built sub-network, successively to sub-network number numbering, by selected just residual point and therewith just residual be connected allly bear residual point and add in newly-built sub-network, using all positive and negative residual point chosen as sub-network node;
(4c) choose the just residual point of the next one, take out and the just residual all negative residual point be connected, whether the negative residual point selected by judgement has added in sub-network all, if so, performs step (4d); Otherwise, perform step (4b);
(4d) judge whether the negative residual point got is connected to the sub-network of same numbering, if so, by got just residual point with bear residual point and add in the sub-network that step (4b) builds, using all positive and negative residual point chosen as sub-network node; Otherwise, perform step (4e);
(4e) sub-network that the numbering at negative residual some place in step (4c) is minimum is taken out, by the just residual point taken out in step (4c), all positive and negative residual point born in the sub-network at residual point and negative residual some place, add in the minimum sub-network of numbering, delete the sub-network at negative residual some place, again to sub-network number consecutively;
(4f) judge that whether just residual point is last just residual point of interferometric phase image, if so, perform step (5); Otherwise, perform step (4c).
(5) ground node is added:
According to the sequence of positions of just residual point, by the border of the interferometric phase image of search in step (3c), add in the sub-network at a just residual place, using the ground node of the border of this interferometric phase image as sub-network.
(6) uneven sub-network is found out:
(6a) judge whether first sub-network reaches the condition of sub-network balance, if so, perform step (6c); Otherwise record does not reach the numbering of the sub-network of equilibrium condition, perform step (6c);
(6b) judge whether next sub-network reaches the condition of sub-network balance, if so, perform step (6c); Otherwise record does not reach the numbering of the sub-network of equilibrium condition;
(6c) judge whether sub-network is last sub-network, if so, perform step (7); Otherwise, perform step (6b).
(7) imbalance correction sub-network:
(7a) according to the order of sub-network numbering, search generates residual some position of each subnet boundaries, the position coordinates of the residual point that record searching arrives; Calculate the Euclidean distance between nearest two the residual points between each sub-network;
(7b) order that the uneven sub-network recorded according to step (6) is numbered, adjacent uneven sub-network is connected between two, generate a unbalance network, with the Euclidean distance between the sub-network in step (7a), the connecting line in this unbalance network is weighted;
(7c) utilize minimum cost flow algorithm, according to the criterion that expense is minimum, ask for the optimal path that in unbalance network, sub-network connects;
(7d) merge the all-ones subnet network on connecting line, build the sub-network that generation one is new.
(8) solvability is judged:
(8a) according to the number order of sub-network, choose first sub-network, according to the criterion that expense is minimum, adopt minimum cost flow algorithm to optimize sub-network; If sub-network exists optimum solution, the optimum solution of record sub-network, performs step (8c); Otherwise, the integer in step (3b) is increased by 10 on original basis, performs step (3c);
(8b) choose next sub-network, according to the criterion that expense is minimum, adopt minimum cost flow algorithm to optimize sub-network; If sub-network exists optimum solution, the optimum solution of record sub-network, performs step (8c); Otherwise, the integer in step (3b) is increased by 10 on original basis, performs step (3c);
(8c) judge whether sub-network is last sub-network, if so, perform step (9); Otherwise, perform step (8b).
(9) branch tangent line is arranged:
According to the optimum solution of each sub-network optimized out, in interferometric phase image, connect the node of optimum solution process, as branch tangent line.
(10) phase gradient integration:
In interferometric phase image, select arbitrarily a pixel as a reference point, do not cross over the mode of arranging branch tangent line according to path of integration, from reference point, integration is carried out to interferometric phase gradient, until whole interferometric phase image integration is complete, obtains solution and twine phase diagram.
The present invention has the following advantages compared with prior art:
First, because the present invention adopts the strategy building sub-network and optimization thereof to process large-scale data, overcoming prior art adopts rectangular partition strategy to split network, and cause the unwrapping phase place between each sub-block to there is discontinuous problem, make the present invention have phase unwrapping around the high advantage of precision.
Second, due to the strategy that the present invention adopts network flow algorithm to be optimized sub-network, overcoming prior art adopts iterative processing to cause time complexity large, and the problem slower to processing speed during large-scale data process, makes the present invention have fireballing advantage.
3rd, because the present invention adopts the strategy be weighted the potential path between residual point, overcoming prior art adopts the shortest optimisation strategy of overall branch tangential length to arrange the problem that a tangent line causes low precision, makes the present invention have the high advantage of the precision of phase unwrapping.
Accompanying drawing explanation
Fig. 1 is process flow diagram of the present invention;
Fig. 2 is simulated effect figure of the present invention.
Embodiment
Below in conjunction with accompanying drawing, the present invention will be further described.
With reference to accompanying drawing 1, specific embodiment of the invention step is as follows:
Step 1, reads data.
The first step, reads interferometric phase diagram data.Interferometric phase diagram data is the phase data be wound around through 2 π generated after interference treatment.
Second step, reads interferometric phase image quality information data.Interferometric phase image quality information data, for retraining the setting of a tangent line, is weighted the path between residual point, and interferometric phase image quality information data can be coefficient of coherence etc.
Step 2, identifies residual point.
Phase gradient between unwrapping algorithm hypothesis neighbor, between-π to π, carries out integration to phase gradient and obtains unwrapping phase place.But by the impact of interferometric phase noise and discontinuous landform, the phase gradient of actual interferometric phase image cannot all meet above-mentioned assumed condition.If the phase gradient information that phase unwrapping make use of these mistakes around Integral Processing will cause phase unwrapping around error, and this error will be accumulation.
Identify that the concrete steps of residual point are:
The first step, by first pixel I in interferometric phase image 1,1position, as the position of extracting pixel-phase value in interferometric phase image; Wherein, I 1,1represent that in interferometric phase image, line number and row number are the pixel of 1;
Second step, extracts four pixel I in interferometric phase image successively n,Mi n, M+1i n+1, Mi n+1, M+1phase value, will pixel I be extracted n+1, Mposition mark be the position of extracting pixel-phase value next time, the phase value of four pixels of extraction generates an arest neighbors closed-loop path; Wherein, I n,Mrepresent the pixel that in interferometric phase image, line number is N, row number are M, I n, M+1represent the pixel that in interferometric phase image, line number is N, row number are M+1, I n+1, Mrepresent the pixel that in interferometric phase image, line number is N+1, row number are M, I n+1, M+1represent the pixel that in interferometric phase image, line number is N+1, row number are M+1;
3rd step, carries out integration along clockwise direction to arest neighbors closed-loop path, if integral result is 2 π, by the pixel I of four in closed-loop path n,Mi n, M+1i n+1, Mi n+1, M+1center in the interferometric phase image of place is labeled as just residual point; If integral result is-2 π, by the pixel I of four in closed-loop path n,Mi n, M+1i n+1, Mi n+1, M+1center in the interferometric phase image of place is labeled as negative residual point;
4th step, judges the pixel I extracted n+1, M+1be whether last pixel of interferometric phase image, if so, perform step 3; Otherwise, perform second step.
Step 3, arranges potential branch tangent line path.
Phase unwrapping is around avoiding not meeting (-π, π] phase gradient of assumed condition carries out Integral Processing, this can intercept path of integration by arranging a tangent line between positive and negative residual point, branch tangent line to act as in balance path of integration positive and negative residual counts out.
Access path in phase diagram between residual point has multiple, does not have concrete connected mode, and can only connect a branch tangent line between actual two residual points, the object of this step chooses rational branch tangent line path.
The concrete steps arranging potential branch tangent line path are:
(3a) integer range is set as by 30 ~ 150;
(3b) for the residual point that density is large, a less integer is chosen in integer range, for the residual point that density is little, in integer range, choose a larger integer, using the integer chosen as the maximum distance between just residual point and negative residual point and with just residual be connected maximumly bear residual counting;
(3c) position of the residual point marked according to step (2c), around just residual at first, the from the close-by examples to those far off border of the negative residual point of search and interferometric phase image; Judge whether the border three of just residual point, negative residual point and interferometric phase image meets to impose a condition, if met, perform step (3e); Otherwise, perform step (3h);
(3d) around the just residual point of the next one, the from the close-by examples to those far off border of the negative residual point of search and interferometric phase image; Judge whether the border three of just residual point, negative residual point and interferometric phase image meets to impose a condition, if met, perform step (3e); Otherwise, perform step (3h);
(3e) by the residual point that just residual point and all negative residual composition corresponding with it match, network to be optimized is formed by the residual of all pairings;
(3f) weighted optimization network to be optimized, chooses optimum access path between residual point.The concrete steps of this step are:
The first step, extracts in network to be optimized and matches the position of residual point;
Second step, in interferometric phase Quality Map on residual corresponding position, extracts and allly can connect the data of pairing residual some institute through node, matching the path of residual point and is weighted, generating the weight path of many to connecting;
3rd step, adopts network flow optimization algorithm, minimum as optiaml ciriterion using weights, selects an optimum access path from many weight paths;
(3g) in interferometric phase image, all nodes of optimal path process are connected, using connecting line as potential branch tangent line path;
(3h) judge whether just residual point is last just residual point, if so, perform step (4); Otherwise, perform step (3d);
Step 4, builds sub-network.
For extensive interference treatment data, the scale of the residual some branch tangent line network utilizing entire image to build is very large, and the efficiency causing minimum cost flow algorithm to solve most brachyplast tangent line sharply declines by this.For raising the efficiency, several sub-networks will be built to entire image and processing.
The concrete steps of building sub-network are:
The first step, according to the sequence of positions of just residual point in interferometric phase image, chooses first just residual point and the just residual all negative residual point be connected therewith;
Second step, a newly-built sub-network, successively to sub-network number numbering, by selected just residual point and therewith just residual be connected allly bear residual point and add in newly-built sub-network, using all positive and negative residual point chosen as sub-network node;
3rd step, chooses the just residual point of the next one, and take out and the just residual all negative residual point be connected, whether the negative residual point selected by judgement has added in sub-network all, if so, performs the 4th step; Otherwise, perform second step;
4th step, judges whether the negative residual point got is connected to the sub-network of same numbering, if so, by got just residual point with bear residual point and add in the sub-network that the 3rd step builds, using all positive and negative residual point chosen as sub-network node; Otherwise, perform the 5th step;
5th step, take out the sub-network that the numbering at negative residual some place in the 3rd step is minimum, by the just residual point taken out in the 3rd step, all positive and negative residual point born in the sub-network at residual point and negative residual some place, add in the minimum sub-network of numbering, delete the sub-network at negative residual some place, again to sub-network number consecutively;
6th step, judges that whether just residual point is last just residual point of interferometric phase image, if so, performs step 5; Otherwise, perform the 3rd step.
Step 5, adds ground node.
Drill in doing, only comprise just residual point and negative residual point in sub-network, and do not comprise the border of interferometric phase image, the namely ground node of sub-network, therefore, need the ground node of positive and negative residual connection to add in respective sub-network.The step of adding ground node is: according to the sequence of positions of just residual point, by the border of the interferometric phase image of three-wave mixing in step 3, add in the sub-network at a just residual place, using the ground node of the border of this interferometric phase image as sub-network;
Step 6, finds out uneven sub-network.
After the residual point of large-scale data is divided into different sub-networks, in sub-network, residual point charge not necessarily balances.Unbalanced sub-network is optimized and there is no optimum solution.
The concrete steps finding out uneven sub-network are:
The first step, judges whether first sub-network reaches the condition of sub-network balance; The condition of sub-network balance is for meeting one of following two conditions, and first condition is containing ground node in sub-network; Second condition is that in sub-network, just residual counting equals negative and residually to count; If so, the 3rd step is performed; Otherwise the numbering of record sub-network, performs the 3rd step;
Second step, judges whether next sub-network reaches the condition of sub-network balance; The condition of sub-network balance is for meeting one of following two conditions, and first condition is containing ground node in sub-network; Second condition is that in sub-network, just residual counting equals negative and residually to count; If so, the 3rd step is performed; Otherwise, the numbering of record sub-network;
3rd step, judges whether sub-network is last sub-network, if so, performs step 7; Otherwise, perform second step.
Step 7, imbalance correction sub-network.
The object of imbalance correction sub-network makes all sub-networks all reach equilibrium state.The concrete steps of imbalance correction are:
The first step, according to the order of sub-network numbering, search generates residual some position of each subnet boundaries, the position coordinates of the residual point that record searching arrives; Calculate the Euclidean distance between nearest two the residual points between each sub-network; The formula calculating Euclidean distance between nearest two residual points is:
D = ( i 1 - i 2 ) 2 + ( j 1 - j 2 ) 2
Wherein, D represents the Euclidean distance between nearest two residual points, i 1and i 2represent the horizontal ordinate of nearest two residual points respectively, j 1and j 2represent the ordinate of nearest two residual points respectively;
Second step, the order of the uneven sub-network numbering recorded according to step 6, connects adjacent uneven sub-network between two, generates a unbalance network, be weighted with the Euclidean distance between the sub-network in the first step to the connecting line in this unbalance network;
3rd step, utilizes minimum cost flow algorithm, according to the criterion that expense is minimum, asks for the optimal path that in unbalance network, sub-network connects;
4th step, merges the all-ones subnet network on connecting line, builds the sub-network that generation one is new.
Step 8, judges solvability.
Obtain sub-network after calibrated is balance on electric charge, but the sub-network obtained not necessarily can be separated, and needs the solvability judging sub-network.
Judge that the concrete steps of sub-network solvability are:
The first step, according to the number order of sub-network, chooses first sub-network, according to the criterion that expense is minimum, adopts minimum cost flow algorithm to optimize sub-network; If sub-network exists optimum solution, the optimum solution of record sub-network, performs the 3rd step; Otherwise, the integer in second step in step 3 is increased by 10 on original basis, performs the 3rd step in step 3;
Second step, chooses next sub-network, according to the criterion that expense is minimum, adopts minimum cost flow algorithm to optimize sub-network; If sub-network exists optimum solution, the optimum solution of record sub-network, performs the 3rd step; Otherwise, the integer in second step in step 3 is increased by 10 on original basis, performs the 3rd step in step 3;
3rd step, judges whether sub-network is last sub-network, if so, performs step 9; Otherwise, perform second step.
Step 9, arranges branch tangent line.
According to the feasible solution of each sub-network asking for out, in interferometric phase image, connect the node of feasible solution process, as branch tangent line.
Step 10, phase gradient integration.
In interferometric phase image, select arbitrarily a pixel as a reference point, do not cross over the mode of arranging branch tangent line according to path of integration, from reference point, integration is carried out to interferometric phase gradient, until whole interferometric phase image integration is complete, obtains solution and twine phase diagram.
Below in conjunction with measured data, effect of the present invention is further described.
The two width heavy rail interference SAR image experiment Analysis that the present invention selects German TerraSAR-X satellite to obtain.Experimental situation of the present invention is common computer, and the concrete configuration of computing machine is i7 processor, 2.5GHz dominant frequency, 16GB internal memory, VisualStudio2010 software.Under above-mentioned experimental situation, method of the present invention and prior art is adopted to carry out phase unwrapping around process to experimental data.
The original echo admission time of two width SAR image of the present invention, be respectively on February 25th, 2010 and on March 8th, 2010, interferometric phase image orientation is 24949 to length, distance is 18182 to length, the original interference phase place of major-minor SAR image is as accompanying drawing 2(a) shown in, residual the counting of interferometric phase image is 1864806.
The present invention adopts coefficient of coherence weighting when arranging potential path, and the coefficient of coherence figure of original interference phase place is as accompanying drawing 2(b) shown in.Accompanying drawing 2(c) for utilizing the minimum cost flow algorithm in the interference treatment software Gamma software of current main-stream, the result figure of phase unwrapping around process is carried out to data; Accompanying drawing 2(d) for the present invention to data carry out phase unwrapping around process result figure.
As can be seen from the upper right corner of Fig. 2 (c), Gamma software minimum cost flow algorithm result, there is obvious bulk zone phase hit, this adopts rectangular partition to cause by Gamma software.Solution twines in phase diagram and occurs phase hit, illustrates that the precision that solution twines is low.From accompanying drawing 2(d), the present invention, by building the strategy of sub-network, efficiently avoid the problem occurring phase hit.Can find out that the result that solution of the present invention twines does not have hop region by Fig. 2 (d), illustrate that the precision that solution of the present invention twines is high.In this experiment, the time adopting the process of prior art Gamma software is 2646s, and the processing time of the present invention is 519s, is contrasted by the time, also can find out that the present invention has fireballing advantage.

Claims (3)

1., based on the phase unwrapping winding method of large-scale data weighting branch tangent line, comprise the steps:
(1) data are inputted:
(1a) the interferometric phase diagram data of input after interference treatment;
(1b) quality information data of interferometric phase image is inputted;
(2) residual point is identified:
(2a) by first pixel I in interferometric phase image 1,1position, as the position of extracting pixel-phase value in interferometric phase image; Wherein, I 1,1represent that in interferometric phase image, line number and row number are the pixel of 1;
(2b) four pixel I in interferometric phase image are extracted successively n,Mi n, M+1i n+1, Mi n+1, M+1phase value, will pixel I be extracted n+1, Mposition mark be the position of extracting pixel-phase value next time, the phase value of four pixels of extraction generates an arest neighbors closed-loop path; Wherein, I n,Mrepresent the pixel that in interferometric phase image, line number is N, row number are M, I n, M+1represent the pixel that in interferometric phase image, line number is N, row number are M+1, I n+1, Mrepresent the pixel that in interferometric phase image, line number is N+1, row number are M, I n+1, M+1represent the pixel that in interferometric phase image, line number is N+1, row number are M+1;
(2c) along clockwise direction integration is carried out to arest neighbors closed-loop path, if integral result is 2 π, by the pixel I of four in closed-loop path n,Mi n, M+1i n+1, Mi n+1, M+1center in the interferometric phase image of place is labeled as just residual point; If integral result is-2 π, by the pixel I of four in closed-loop path n,Mi n, M+1i n+1, Mi n+1, M+1center in the interferometric phase image of place is labeled as negative residual point;
(2d) the pixel I extracted is judged n+1, M+1be whether last pixel of interferometric phase image, if so, perform step (3); Otherwise, perform step (2b);
(3) potential branch tangent line path is set:
(3a) integer range is set as by 30 ~ 150;
(3b) for the residual point that density is large, a less integer is chosen in integer range, for the residual point that density is little, in integer range, choose a larger integer, using the integer chosen as the maximum distance between just residual point and negative residual point and with just residual be connected maximumly bear residual counting;
(3c) position of the residual point marked according to step (2c), around just residual at first, the from the close-by examples to those far off border of the negative residual point of search and interferometric phase image; Judge whether the border three of just residual point, negative residual point and interferometric phase image meets to impose a condition, if met, perform step (3e); Otherwise, perform step (3h);
Described imposing a condition refers to, meets the condition that namely one of following two conditions meet setting:
The first, the distance between the border of negative residual point and interferometric phase image and just residual point is greater than the integer selected by step (3b);
The second, the negative residual integer be greater than selected by step (3b) of counting be connected with just residual point;
(3d) around the just residual point of the next one, the from the close-by examples to those far off border of the negative residual point of search and interferometric phase image; Judge whether the border three of just residual point, negative residual point and interferometric phase image meets to impose a condition, if met, perform step (3e); Otherwise, perform step (3h);
Described imposing a condition refers to, meets the condition that namely one of following two conditions meet setting:
The first, the distance between the border of negative residual point and interferometric phase image and just residual point is greater than the integer selected by step (3b);
The second, the negative residual integer be greater than selected by step (3b) of counting be connected with just residual point;
(3e) by the residual point that just residual point and all negative residual composition corresponding with it match, network to be optimized is formed by the residual of all pairings;
(3f) weighted optimization network to be optimized, chooses optimum access path between residual point;
(3g) in interferometric phase image, all nodes of optimal path process are connected, using connecting line as potential branch tangent line path;
(3h) judge whether just residual point is last just residual point, if so, perform step (4); Otherwise, perform step (3d);
(4) sub-network is built:
(4a) according to the sequence of positions of just residual point in interferometric phase image, first just residual point and the just residual all negative residual point be connected therewith is chosen;
(4b) a newly-built sub-network, successively to sub-network number numbering, by selected just residual point and therewith just residual be connected allly bear residual point and add in newly-built sub-network, using all positive and negative residual point chosen as sub-network node;
(4c) choose the just residual point of the next one, take out and the just residual all negative residual point be connected, whether the negative residual point selected by judgement has added in sub-network all, if so, performs step (4d); Otherwise, perform step (4b);
(4d) judge whether the negative residual point got is connected to the sub-network of same numbering, if, got just residual point and negative residual point are added in the sub-network that step (4b) builds, using all positive and negative residual point chosen as sub-network node, perform step (4f); Otherwise, perform step (4e);
(4e) sub-network that the numbering at negative residual some place in step (4c) is minimum is taken out, by the just residual point taken out in step (4c), all positive and negative residual point born in the sub-network at residual point and negative residual some place, add in the minimum sub-network of numbering, delete the sub-network at negative residual some place, again to sub-network number consecutively;
(4f) judge that whether just residual point is last just residual point of interferometric phase image, if so, perform step (5); Otherwise, perform step (4c);
(5) ground node is added:
According to the sequence of positions of just residual point, by the border of the interferometric phase image of search in step (3c), add in the sub-network at a just residual place, using the ground node of the border of this interferometric phase image as sub-network;
(6) uneven sub-network is found out:
(6a) judge whether first sub-network reaches the condition of sub-network balance, if so, perform step (6c); Otherwise record does not reach the numbering of the sub-network of equilibrium condition, perform step (6c);
Described sub-network equilibrium condition refers to, meets one of following two conditions and namely meets sub-network equilibrium condition:
The first, containing ground node in sub-network;
The second, in sub-network, just residual counting equals negative and residually to count;
(6b) judge whether next sub-network reaches the condition of sub-network balance, if so, perform step (6c); Otherwise record does not reach the numbering of the sub-network of equilibrium condition;
Described sub-network equilibrium condition refers to, meets one of following two conditions and namely meets sub-network equilibrium condition:
The first, containing ground node in sub-network;
The second, in sub-network, just residual counting equals negative and residually to count;
(6c) judge whether sub-network is last sub-network, if so, perform step (7); Otherwise, perform step (6b);
(7) imbalance correction sub-network:
(7a) according to the order of sub-network numbering, search generates residual some position of each subnet boundaries, the position coordinates of the residual point that record searching arrives; Calculate the Euclidean distance between nearest two the residual points between each sub-network;
(7b) order that the uneven sub-network recorded according to step (6) is numbered, adjacent uneven sub-network is connected between two, generate a unbalance network, with the Euclidean distance between the sub-network in step (7a), the connecting line in this unbalance network is weighted;
(7c) utilize minimum cost flow algorithm, according to the criterion that expense is minimum, ask for the optimal path that in unbalance network, sub-network connects;
(7d) merge the all-ones subnet network on connecting line, build the sub-network that generation one is new;
(8) solvability is judged:
(8a) according to the number order of sub-network, choose first sub-network, according to the criterion that expense is minimum, adopt minimum cost flow algorithm to optimize sub-network; If sub-network exists optimum solution, the optimum solution of record sub-network, performs step (8c); Otherwise, the integer in step (3b) is increased by 10 on original basis, performs step (3c);
(8b) choose next sub-network, according to the criterion that expense is minimum, adopt minimum cost flow algorithm to optimize sub-network; If sub-network exists optimum solution, the optimum solution of record sub-network, performs step (8c); Otherwise, the integer in step (3b) is increased by 10 on original basis, performs step (3c);
(8c) judge whether sub-network is last sub-network, if so, perform step (9); Otherwise, perform step (8b);
(9) branch tangent line is arranged:
According to the optimum solution of each sub-network optimized out, in interferometric phase image, connect the node of optimum solution process, as branch tangent line;
(10) phase gradient integration:
In interferometric phase image, select arbitrarily a pixel as a reference point, do not cross over the mode of arranging branch tangent line according to path of integration, from reference point, integration is carried out to interferometric phase gradient, until whole interferometric phase image integration is complete, obtains solution and twine phase diagram.
2. the phase unwrapping winding method based on large-scale data weighting branch tangent line according to claim 1, is characterized in that: the weighted optimization network to be optimized in step (3f), and the concrete steps choosing optimum access path are as follows:
The first step, extracts in network to be optimized and matches the position of residual point;
Second step, in interferometric phase Quality Map on residual corresponding position, extracts and allly can connect the data of pairing residual some institute through node, matching the path of residual point and is weighted, generating the weight path of many to connecting;
3rd step, adopts network flow optimization algorithm, minimum as optiaml ciriterion using weights, selects an optimum access path from many weight paths.
3. the phase unwrapping winding method based on large-scale data weighting branch tangent line according to claim 1, is characterized in that: between nearest two the residual points of the calculating described in step (7a), the formula of Euclidean distance is:
D = ( i 1 - i 2 ) 2 + ( j 1 - j 2 ) 2
Wherein, D represents the Euclidean distance between nearest two residual points, i 1and i 2represent the horizontal ordinate of nearest two residual points respectively, j 1and j 2represent the ordinate of nearest two residual points respectively.
CN201410105328.XA 2014-03-20 2014-03-20 Based on the phase unwrapping winding method of large-scale data weighting branch tangent line Active CN103869290B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201410105328.XA CN103869290B (en) 2014-03-20 2014-03-20 Based on the phase unwrapping winding method of large-scale data weighting branch tangent line

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201410105328.XA CN103869290B (en) 2014-03-20 2014-03-20 Based on the phase unwrapping winding method of large-scale data weighting branch tangent line

Publications (2)

Publication Number Publication Date
CN103869290A CN103869290A (en) 2014-06-18
CN103869290B true CN103869290B (en) 2016-03-02

Family

ID=50908029

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201410105328.XA Active CN103869290B (en) 2014-03-20 2014-03-20 Based on the phase unwrapping winding method of large-scale data weighting branch tangent line

Country Status (1)

Country Link
CN (1) CN103869290B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109886880B (en) * 2019-01-03 2021-02-26 杭州电子科技大学 Optical image phase unwrapping method based on U-Net segmentation network

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102621549A (en) * 2011-10-14 2012-08-01 中国人民解放军国防科学技术大学 Multi-baseline/multi-frequency-band interference phase unwrapping frequency domain quick algorithm
CN103279945A (en) * 2013-04-26 2013-09-04 北京理工大学 Quality map guide method and branch cutting method-based interferogram unwrapping method

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102621549A (en) * 2011-10-14 2012-08-01 中国人民解放军国防科学技术大学 Multi-baseline/multi-frequency-band interference phase unwrapping frequency domain quick algorithm
CN103279945A (en) * 2013-04-26 2013-09-04 北京理工大学 Quality map guide method and branch cutting method-based interferogram unwrapping method

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
Residue vector, an approach to branch-cut placement in phase unwrapping: theoretical study;Salah A. Karout et al.;《APPLIED OPTICS》;20070720;第46卷(第21期);4712-4727 *
Robust two-dimensional weighted and unweighted phase unwrapping that uses fast transforms and iterative methods;Dennis C. Ghiglia et al.;《Optical Society of America》;19940131;第11卷(第1期);107-117 *
相位解缠枝切法中的枝切线设置的优化;张昆 等;《微计算机信息》;20080531;第24卷(第5-1期);290-291,294 *

Also Published As

Publication number Publication date
CN103869290A (en) 2014-06-18

Similar Documents

Publication Publication Date Title
CN103902004B (en) Charging remaining time calculation method and device and mobile equipment
US9212917B2 (en) Method and device for matching public transportation route with road network
CN109246592A (en) A kind of acquisition methods and device of position information of user's terminal
CN108280866A (en) Road Processing Method of Point-clouds and system
Komm et al. Treasure hunt with advice
CN105813172B (en) A kind of LTE slave pattern searches the method and system of net
CN104538968A (en) Method and device for determining voltage reactive sensitivity of automatic voltage control system
CN103747419A (en) Indoor positioning method based on signal intensity difference values and dynamic linear interpolation
CN102339478B (en) Method and device for generating digital elevation model from contour map
CN103984822A (en) Three phase flow implementation method based on advanced digital power system simulator
CN108307339A (en) User terminal localization method, system, electronic equipment and storage medium
CN103500168A (en) Method and system for discovering communities in overlapped complex networks according to topology potential
CN103869290B (en) Based on the phase unwrapping winding method of large-scale data weighting branch tangent line
CN105828393A (en) Method and device for inter-system neighbor cell configuration
CN104080169B (en) A kind of underwater wireless sensor network dynamic self-adapting localization method
CN103957544A (en) Method for improving survivability of wireless sensor network
CN106157371A (en) The efficient gridding method of dispersion point cloud based on self adaptation adjacent circumferential expansion strategy
CN104778260B (en) A kind of dynamic radar environmental knowledge storehouse modeling method
CN109781126A (en) A kind of global static path planning method based on vector method
CN109412149A (en) Power grid subgraph construction method, topology analyzing method and device based on region division
CN109255433A (en) A method of community's detection based on similitude
CN105491574A (en) Method and device for determining priority of neighbor regions in neighbor region planning
Bouchoucha et al. Finding link topology of large scale networks from anchored hop count reports
CN106779091A (en) A kind of periodic vibration signal localization method based on transfinite learning machine and arrival distance
CN109946721A (en) Mirror point removing method and system in a kind of satellite positioning

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant