An Integrated Image Classification Approach to Detect the Flood Prone Areas using Sentinel-1 Images
2023 10th International Conference on Computing for Sustainable Global Development (INDIACom)(2023)
摘要
The untimely monsoon floods and the rise in the level of rivers in the north Indian state of Uttar Pradesh (UP) is a major challenge for coordinating relief and rescue operations. To address this challenge, the experts that are deploying a variety of techniques to evaluate the damages might get useful information from synthetic aperture radar (SAR) images which offer images in close to real-time. The Sentinel-1A satellite images from the European Space Agency provide C-band SAR images that have helped researchers achieve enhanced processes of flood detection and flood mapping in flood-prone areas. In regular cases, the Random Forest (RF) and K-Nearest Neighbor (KNN) supervised classification techniques provide higher accuracies than the unsupervised K-Means classification techniques. The integrated approach with the resultant accuracy provides long-reliable elucidation towards flood damages and assessment control. Thenceforth, the RF and KNN supervised classifications based outcomes were integrated with the outcomes of K-Means iterative clustering unsupervised classification producing an enhanced overall accuracy. The overall accuracy from the integrated approach for the city of Basti and Ayodhya is 97% and 93.8% which is higher than the accuracies derived from the supervised and unsupervised classifications. Further, these processed results when refined by time-series analysis over a number of years along with implementing change detection techniques would help in monitoring the flood situation and undertaking precautionary measures.
更多查看译文
关键词
Unsupervised and Supervised Classification,Sentinel-1A,Flood mapping,Clustering,SAR images
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要