A kind of InSAR same place automatic selection of optimal method based on multi-source information association
Technical field
The invention belongs to InSAR terrain mapping technology, relate to a kind of InSAR same place automatic selection of optimal method based on multi-source information association.
Background technology
When the SAR data scaling of Duo Tiao air strips, large region or splicing, tend to run into the inconsistent problem of three-dimensional coordinate of overlapping region.Can be by reference mark information to image calibration one by one, but need a large amount of reference mark, this surveys district for difficulty, is difficult to realize.By the area adjustment thought in photogrammetric, utilize a small amount of reference mark and same place, can set up the restriction relation between each scape image, thereby solve under the condition of sparse reference mark, combine Scaling Problem.By image processing method, can obtain a large amount of same places in overlapping region, many air strips, because the distortion of SAR oblique distance image geometry is serious, make some same place there will be position deviation, same place extractive technique is many simultaneously removes to evaluate same place quality good or not from image process angle, for InSAR, atural object characteristic and the physical meaning of same place representative is more important, therefore how from a large amount of same places, to filter out the useful same place of InSAR is become to a new problem.
By the geometric properties conversion between same place, can identify the point that error is larger, yet even if the same place position remaining is all correct, can not be directly used in and interferes calibration.In InSAR, the interferometric phase of same place present position will be for calibration, and interferometric phase has strict physical meaning, and it is the phase differential that relevant electromagnetic wave produces at same place place, as very weak to reflection of electromagnetic wave ability in river, lake due to water body, almost can not get correct phase differential.In addition because being actually of receiving of radar is wound around into [π, π] between phase place, must recover true value by phase unwrapping method, when same place is on trees, trees volume scattering effect plays a leading role, the phase place that solution winds out can not accurately represent the height of tree, when same place is on house, due to the distinctive folded effect of covering of SAR, make phase value herein cannot recover true value, when same place is in shadow region, because radar cannot receive any echo, can not get phase place at all.Finally, for interfering calibration, use as far as possible level land point under equal conditions, same place extracts can introduce site error, and due in the larger region of topographic relief, site error has amplification to vertical error, so level land point can effectively suppress this impact.
Above-mentioned these problems, can automatically identify by artificial or image processing method, and for the problem of being combined with real physical characteristics, simple image processing method have been difficult to, and need to find a kind of same place method for optimizing for InSAR for positional fault.
Summary of the invention
The object of this invention is to provide a kind of InSAR same place automatic selection of optimal method based on multi-source information association, to solve area adjustment thought, be applied to the same place On The Choice that InSAR produces.
For achieving the above object, technical solution of the present invention is as follows:
Be somebody's turn to do the InSAR same place automatic selection of optimal method based on multi-source information association, comprise the following steps:
Step 1: calculate the coefficient of coherence of the two scape coefficient of coherence figure that same place is corresponding, if having a corresponding coefficient of coherence of point in same place lower than threshold value, remove this group same place;
Step 2: by remaining same place after step 1 is rejected, calculate the bright spot proportion under its window in two scape map of magnitudes, removal ratio is greater than the same place of preset value;
Step 3: by remaining same place after step 2 is rejected, calculate its gray level co-occurrence matrixes in corresponding two scape magnitude image, select this statistic of correlativity, remove the same place that correlativity is greater than threshold value T1;
Step 4: calculation procedure 3 is rejected same place standard deviation under window in two scape DEM afterwards, removes the same place that standard deviation is greater than threshold value T2;
Step 5: if residue same place quantity falls into destination number interval, same place is preferably complete; When residue same place quantity is less than the minimum value in destination number interval, increase threshold value T1, T2, return to step 3 and continue down to carry out; When residue same place quantity is greater than the maximal value in destination number interval, reduce threshold value T1, T2, return to step 3 and continue down to carry out.
Beneficial effect of the present invention: the present invention utilizes image processing method to distinguish the residing type of ground objects of same place, in conjunction with the Physics eigenvector in InSAR, filter out to meet simultaneously and interfere the needed same place of calibration, effectively suppress to interfere the vertical error of calibration to be introduced.
Embodiment
An InSAR same place automatic selection of optimal method for multi-source information association, embodiment is as follows:
Step 1: the coefficient of coherence that calculates the two scape coefficient of coherence figure that same place is corresponding, if have a corresponding coefficient of coherence of point in same place lower than threshold value, be set to 0.4, remove this group same place, what this step was mainly removed is water body and the same place above shade.
Step 2: by remaining same place after step 1 is rejected, calculate the bright spot proportion under its certain window 101*101 in two scape map of magnitudes, 1/2 the point that definition amplitude is greater than maximum amplitude is bright spot.According to bright spot proportion, if ratio is greater than 0.5, remove this same place.This step is mainly removed near the same place strong scattering points such as house.
Step 3: by remaining same place after step 2 is rejected, calculate its gray level co-occurrence matrixes in corresponding two scape magnitude image, select this statistic of correlativity, remove the same place that correlativity is greater than threshold value T1, initial T1 gets 0.1.
While calculating gray level co-occurrence matrixes, window is set to 101*101, and angle is set to 0 ° (horizontal direction), and pixel distance is set to 1, and grey level quantization grade is 16.14 kinds of characteristic statistics that can obtain commonly using from gray level co-occurrence matrixes, select correlativity as statistic here.
Related definition is:
Wherein
Wherein L is the gray shade scale after quantizing.
The ASSOCIATE STATISTICS amount of calculating according to the gray level co-occurrence matrixes of each point, compares itself and threshold value T1, is initially set to empirical value.This step is mainly removed the point on trees.
Step 4: calculation procedure 3 is rejected same place afterwards standard deviation under certain window in two scape DEM, removes the same place that standard deviation is greater than threshold value T2.Wherein, the calculation window of standard deviation is identical with the window that calculates gray level co-occurrence matrixes, and threshold value T2 is initially set to empirical value 1.5m.This step is mainly removed the point that the first-class ratio of slope in mountain region is larger.
Step 5: if residue same place quantity falls into destination number interval (getting 40-60), same place is preferably complete; When residue same place quantity is less than the minimum value in destination number interval, increase threshold value T1, T2, return to step 3 and continue down to carry out; When residue same place quantity is greater than the maximal value in destination number interval, reduce threshold value T1, T2, return to step 3 and continue down to carry out.
Wherein the increase of T1, T2 and minimizing are independently, and such as the each change of T2 is set to 0.05m, and T1 is set to 0.01 at every turn, the interval of T1 [0.05,0.15], the interval of T2 is [0,2],, if maximum (little) value goes beyond the scope, with interval maximum (little) value, substitute.