Research On Layover And Shadow Detecting In InSAR

Synthetic aperture radar interferometry( InSAR) is an increasingly expanding technique allowing for the estimation of three-dimensional terrain images with highly spatial resolution and high accuracy.InSAR needs several observations over the same scene. And a digital elevation model(DEM) or topography-chaging map can be converted by the knowledge of the interferometric geometry and phase differences between each pair of the received signals.The wide applications of InSAR in many areas such as military and scientific researches insure that it has been one of the most active fields in remote sensing.How to detect layover and shadow from the pair of images is one of the difficulties in InSAR processing. There are often some regions where several echos mixed together or no radar echo received in InSAR images and the former is called shadow, the latter layover. Both layover and shadow will destroy the continuity of InSAR phase diagram, and put obstacle to the subsequent phase unwrapping, which can not be avoided completely. Within the background of InSAR system, the dissertation studies the detection of layover and shadow based on single-baseline InSAR. The main contributions of the thesis are summarized in the following.Firstly, based on the mathematical model of interferometric phase, the components of the interferometric phase are deduced as math expressions. Meanwhile, the sensitivity of InSAR measuring including the relationship between interferometric phase and ground elevation is attracted, which is verified by the use of the actual parameters of ERS-1/2.Secondly, in the research of the geometry model of InSAR imaging, the distribution of the multiple coherence coefficient and amplitude among the different image regions indicate a large enough difference,which is used to detect layover and shadow. Experiments on simulated data showed the proposed method can be used to detect layover and shadow in a high-efficiency.Thirdly, in the research of the signal model of InSAR imaging, the phase distortion in layover and shadow leads to great difficulty in phase unwrapping. Further discussion on the auto-covariance matrix of interferometric signal indicates that the eigenvalue clustering of the auto-covariance matrix is closely linked to layover and shadow. Based on that, the thesis proposed a new constructing approach of the joint auto-covariance matrix to detect layover and shadow, which can ensure the effectiveness and reliability in the further InSAR processing.Finally, based on the statistical learning theory, the probability density function of normal InSAR fields estimated by the Gaussian kernel weight function is highlighted. The paper proposed a CFAR algorithm for layover and shadow detection by applying the kernel density estimation for the statistical modeling of InSAR images. Experiments on both simulated and real InSAR data showed the proposed method can be used for layover and shadow detection efficiently.