Information Extraction And Application Of PolSAR Data Based On Special Unitary Groups

Compared with the traditional monopolarized synthetic aperture radar(SAR),the polarimetric SAR(PolSAR)system is capable of acquiring the fully polarimetric scattering information about the observed targets(both manmade target and natural distributed target)thus can identify and analyze them more comprehensively.With the continuous development of the microwave remote sensing technology,PolSAR as one of the research hotspots in the field of remote sensing is widely applied to target recognition,land-use classification,disaster monitoring,terrain slope estimation,etc.Polarimetric decomposition is an indispensable approach in analyzing and extracting target scattering characteristics.Although there is a huge amount of decomposition algorithms proposed for various targets and scenes,there are still two vital issues on polarimetric decomposition that need to be addressed.First,the physical significance of some of the extracted polarimetric parameters are unclear hitherto.Besides,the exploited information by most of the algorithms do not match the degree of freedom(Do F)of the target scattering information,i.e.,the number of the extracted parameters and the Do F are unequal,which leads to information loss or parameter redundancy.Accordingly,it can be inferred that based on a complete and physicalreasonable parameter extraction,with specific representation of the scattering characteristics,the target can be interpreted comprehensively thus an accurate result can be achieved for further target recognition and land-use classification.Based on the theory of polarimetric algebra,the special unitary(SU)group has been employed in simplifying the representation of the polarimetric state of the electromagnetic wave on one hand,and on the other hand,the SU group also shows its efficacy in describing the polarimetric information of the target.There are many advantages of the polarimetric decomposition by using the SU group,i.e.,the decomposition of the scattering matrix is unique,the extraction of the polarimetric information is complete,and the extracted parameters have specific physical significance,which is also the reason for this dissertation focusing on the research of“information extraction and application of the PolSAR data based on the SU group”.Considering the information contained by a single target and that by a distributed target have different Do Fs,which will be decomposed by SU groups with different orders,the two kinds of targets are studied separately.Furthermore,as the most widely applied unitary transformation among the SU group,the orientation transformation extracts the common parameter polarization orientation angle(POA)of both the single and the distributed target with clear physical significance,and this part will be studied independently as it contains a lot of contents.Consequently,the major work and innovation points of this dissertation is summarized as follows:1.Study on the target orientation transformation and the correspondingly extracted parameter POA.By modeling the target scattering characteristics on its zero-POA state and proposing the specific geometric definition of the POA,we unified the various existing POA estimation algorithms.Besides,according to the unification theory,we proposed a novel algorithm for terrain POA estimation based on terrain scattering characteristics and its geometric POA definition,which has been shown to be able to solve the range wrapping and estimation bias problems of the existing algorithms effectively.2.Study on the application of second-order SU(SU(2))group in single target polarimetric decomposition and parameter extraction.This part mainly focuses on the existing SU(2)algorithm Huynen-Euler decomposition,and by analyzing and verifying the drawbacks of its extracted parameters,i.e.range compression and angle ambiguity,we proposed new parameter models to improve the identification and classification abilities of the single target SU(2)decomposition algorithm.Furthermore,we presented a fast decomposition approach for SU(2)decomposition,which extracts the polarimetric parameters simultaneously with high computational efficiency.Thus,there is an outstanding advantage of the SU(2)decomposition algorithm in disaster monitoring and urban damage level estimation,not only because of the clear and specific physical significance of its extractions but also for its computational efficiency which can rapidly respond to suddenly outbroken natural disasters,e.g.earthquake and tsunami,etc.3.Study on the application of third-order SU(SU(3))group in distributed target decomposition and parameter extraction.We extended the single target SU(2)decomposition algorithm to the general case of distributed targets based on the SU(3)group with nine parameters totally containing the obtained target polarimetric information so as to extract the information sufficiently and losslessly.Experimental results show that the extractions have a nice performance in terrain classification which reflect the target characteristic clearly and unambiguously.Besides,for the PolSAR data of the distributed target,we also studied the pre-processing algorithm and developed a new visualization method based on channel coherence.