最上等的
计算机科学
地形
合成孔径雷达
连贯性(哲学赌博策略)
干涉合成孔径雷达
时间序列
遥感
算法
人工智能
数学
统计
地质学
地理
机器学习
方位角
地图学
几何学
作者
Zhangfeng Ma,Mi Jiang,Teng Huang
出处
期刊:IEEE Transactions on Geoscience and Remote Sensing
[Institute of Electrical and Electronics Engineers]
日期:2021-06-01
卷期号:59 (6): 4818-4826
被引量:12
标识
DOI:10.1109/tgrs.2020.3009996
摘要
In the coming era of synthetic aperture radar (SAR) big data, the in-orbit Sentinel-1 mission will provide unprecedented data with an increasing volume. As a fundamental step of time-series analysis for such large and growing amount data, terrain observation by progressive scans (TOPS) co-registration still presents a relative challenge: 1) low coherence scenarios may degrade the estimate accuracy and 2) unprecedented and growing data volume increases the computational burden. To overcome both limitations, this article presents a sequential approach for TOPS time-series co-registration, with an emphasis on the enhanced spectral diversity (ESD) estimate accuracy over low coherence scenes. We first employ double sample over the burst overlap region to improve the statistical proprieties of sample covariance matrix, followed by ESD phase estimation using a phase linking algorithm. Then, we carry out the sequential co-registration on each mini-stack without the necessity for reprocessing the entire stack by introducing a data compression technique. Using synthetic data and real Sentinel-1 TOPS data over densely vegetated areas in the Yunan-Kweichow plateau, we fully evaluate the performance of presented approach and compare the results with those obtained from the state-of-the-art techniques. We found that the sequential approach can provide better time-series co-registration accuracy over low coherence scenes with the moderate computational efficiency.
科研通智能强力驱动
Strongly Powered by AbleSci AI