中文摘要：

针对特大地震和海啸灾害的特点与灾情监测的需求，设计和构建了一种基于多特征决策级融合的灾情快速变化检测方法。首先利用遥感影像中提取的NDVI、主成分变换分量、独立成分变换分量等特征分别提取变化信息。然后对多种变化信息进行决策级融合。获得具有更高可靠性的变化图，用于灾情分析。将所设计的方法用于国产环境与灾害监测预报小卫星HJ-1A／B数据处理分析，对日本东部沿海区域海啸灾害前后进行变化检测与灾情信息提取试验，有效地检测了海啸灾害后的海水倒灌区域、陆地积水区、植被淹没区以及建筑受损区等变化区域。研究表明，基于多源特征融合的非监督变化检测流程可以快速、有效地提取海啸受灾区域。为灾害应急响应与灾情评估提供支持。

英文摘要：

The huge tsunami triggered by an earthquake of magnitude 9.0 on March 11, 2011 hit the east coast of Honshu, Japan, and caused serious social and economic losses. According to the characteristics of earthquake and tsunami disaster and the requirements of damage monitoring, a rapid disaster development detection process based on a decision level fusion of multiple features -is designed. In this approach, each feature extracted from the original remote sensing images, including NDVI, NDWI, components of the principal component analysis and the independent component analysis, is used to derive a specific change map, and different change maps are then integrated by a decision level fusion algorithm to generate a synthetic change map with a higher reliability, which can be used for the damage assessment. Multi-temporal HJ-1A/B （environment and disaster monitoring and forecasting of small satellite constellation） images are processed by the proposed approach and used for detecting the devastated areas in east coast of Japan before and after tsunami. The experimental results confirm the feasibility and effectiveness of the proposed approach, and demonstrate the advantages of HJ-1A/B remote sensing data. This unsupervised change detection process can identify the tsunami-devastated regions quickly and efficiently, and provide the technical and decision support for the disaster emergency response and loss evaluation.