4:15 PM - 4:30 PM
[ACG37-21] Using Pleiades and A Semi-Automatic Change Detection Algorithm for Hazard Mapping
Keywords:Simple Linear Iterative Clustering, Gray-Level Co-occurrence Matrix, Land Cover Change
Hazard mapping is crucial for a rapid damage assessment and an overview of the area of effect. It also provides information to support various decision-making procedures and to understand the triggering mechanism behind the natural hazard. To achieve this goal, detecting land cover changes by satellite image that repeatedly captures surface information is a feasible approach. This study aims to develop a system with a semi-automatic algorithm to classify the changed/unchanged areas and produce a binarized map. We adopt the Pleiades images and a direct comparison method based on the object-oriented (OO) approach. The OO approach first applies histogram matching and pan-sharpening to the Pleiades images and segments the pan-sharpening image into parcel levels based on Simple Linear Iterative Clustering (SLIC). Next, the spectral and textural features were extracted on the parcel level. The textural features are extracted by the Grey Level Co-occurrence Matrix (GLCM) indices and the statistic of spectral are computed. Finally, semi-automatic detection is conducted by analyzing the changes in the labeled dataset. Combining textural and spectral information from satellite images, our preliminary results exemplified by an earthquake that occurred on September 18, 2022, in Hualien, Taiwan shows a map that highlights the event-driven change and agrees well with visual detection.