报告题目:Continental-Scale Crop Type Identification and Mapping based on Dual-polarimetric Sentinel-1 Time Series
报告时间:2018年10月17日(星期三)10:00-11:00
报告地点:机电学院二楼会议室
报告人:澳大利亚联邦科学与工业研究组织周正舒教授
报告人简介:
周正舒是澳大利亚联邦科学与工业研究组织(CSIRO)教授,IEEE高级会员,CSIRO遥感和图像分析团队雷达遥感监测小组负责人。于2005年移居澳大利亚,并与阿得雷德大学电气与电子工程学院合作,曾建立了澳大利亚首个用于植被监测和参数提取的地基极化干涉雷达系统,目前主要负责SAR相关技术在农业和陆地监测等领域的工作。自2009年加入CSIRO以来,一直负责大部分雷达相关的项目。研究领域包括雷达系统构建与校正、雷达成像与雷达信号处理、极化SAR干涉测量等。
报告内容简介:
Synthetic Aperture Radar (SAR) provides all day and night imaging capability due to the unique responses of terrain and targets to radar frequencies and the minimum constraints on time-of-day and atmospheric conditions. With the advancements of radar systems and signal processing techniques, radar remote sensing is increasingly significant and available for Earth observation. Since the 1990’s CSIRO have been involved with SAR applications, making use of data from most SAR missions.Due to the recent maturing of the SAR data acquisitions and availability of freely accessible SAR imagery,CSIRO have invested to boost and enhance its SAR capabilities across multiple application domains.
In this talk we give an overview of recent CSIRO work in SAR technique innovations and applications, especially our current work on continental scale crop monitoring. We have foreseen significant growth in the SAR applications for pilot studies as well as developing monitoring systems at continental scale for precision agriculture including cropping and pasture, and other applications. Our practice of crop mapping for Australia’s Wheatbelts is introduced. Evidences of how polarimetric decomposition improving the accuracy of crop type classification and quantitative analysis of crop classification accuracy with various lengths of Sentinel-1 time series are presented.
由于地形和目标对雷达频率的独特响应以及时间和大气条件的微小限制,合成孔径雷达(SAR)提供了全天候成像能力。随着雷达系统和信号处理技术的进步,雷达遥感在对地观测活动中变得越来越重要了。自1990年以来,CSIRO一直利用多种SAR传感器的数据进行SAR应用研究与开发。由于近期SAR数据获取技术的进步以及大量全覆盖、多时相的免费Sentinel-1雷达数据,CSIRO已加强提升在多个SAR应用领域的研究与开发能力。
在本报告中,我们简要介绍最近CSIRO在SAR技术创新以及应用方面的工作,特别是我们目前进行大规模农业遥感系统开发的进展。随着SAR应用试点研究的显著增长,我们预研并开发了在全国范围内的精准农业遥感系统,包括农作物制图和牧草监测以及其它方面的应用。我们集中介绍澳大利亚小麦带产区的作物种类识别与作物制图,并给出如何利用极化分解技术提高作物种类分类准确性和不同长度的Sentinel-1时间序列对作物分类精度影响的定量分析证据。