TY - JOUR
T1 - Combination of super-resolution psi and traditional psi by identification of homogeneous areas
AU - Zhang, Hao
AU - López-Dekker, Paco
AU - Li, Shaoning
PY - 2020
Y1 - 2020
N2 - The performance of Persistent Scatterer Interferometry (PSI) depends heavily on Persistent Scatterer (PS) density. In order to increase PS density, we can apply Super-Resolution reprocessing algorithms in PSI. Involving the reprocessing algorithms and the peak-detection-based Persistent Scatterer Candidate points (PSCs) selection method, the full PSI chain is referred to as Super-Resolution PSI (SR-PSI). The implementation of the Super-Resolution reprocessing algorithm, however, is computationally intensive, which makes SR-PSI time-consuming. In this work, we propose to improve the efficiency by constraining the Capon-based reprocessing to the non-homogeneous areas (e.g., urban areas). We notice that the Capon algorithm performs similarly as the Fourier-based algorithm for homogeneous regions (e.g., grassland), thus we can use Single Look Complex (SLC) images for these areas. With the Coefficient of Variation (CV) as the index, we divide the full image into two classes: homogeneous areas, for which we select PSCs from the original stack, and non-homogeneous areas, for which we extract PSCs from the Capon-based reprocessed images. Then we combine the PSCs of both cases for further PSI processing. We applied the combination method to a stack of TerraSAR-X data. The results show that the proposed approach is more computationally efficient than the original SR-PSI with the effectiveness uncompromised, especially for applications aiming at the urban deformation.
AB - The performance of Persistent Scatterer Interferometry (PSI) depends heavily on Persistent Scatterer (PS) density. In order to increase PS density, we can apply Super-Resolution reprocessing algorithms in PSI. Involving the reprocessing algorithms and the peak-detection-based Persistent Scatterer Candidate points (PSCs) selection method, the full PSI chain is referred to as Super-Resolution PSI (SR-PSI). The implementation of the Super-Resolution reprocessing algorithm, however, is computationally intensive, which makes SR-PSI time-consuming. In this work, we propose to improve the efficiency by constraining the Capon-based reprocessing to the non-homogeneous areas (e.g., urban areas). We notice that the Capon algorithm performs similarly as the Fourier-based algorithm for homogeneous regions (e.g., grassland), thus we can use Single Look Complex (SLC) images for these areas. With the Coefficient of Variation (CV) as the index, we divide the full image into two classes: homogeneous areas, for which we select PSCs from the original stack, and non-homogeneous areas, for which we extract PSCs from the Capon-based reprocessed images. Then we combine the PSCs of both cases for further PSI processing. We applied the combination method to a stack of TerraSAR-X data. The results show that the proposed approach is more computationally efficient than the original SR-PSI with the effectiveness uncompromised, especially for applications aiming at the urban deformation.
KW - Homogeneous area
KW - PSI
KW - SAR
KW - Super-resolution
UR - http://www.scopus.com/inward/record.url?scp=85102825498&partnerID=8YFLogxK
U2 - 10.1109/ACCESS.2020.3028491
DO - 10.1109/ACCESS.2020.3028491
M3 - Article
AN - SCOPUS:85102825498
SN - 2169-3536
VL - 8
SP - 181640
EP - 181649
JO - IEEE Access
JF - IEEE Access
ER -