Hybrid deep learning approach for pre-flood and post-flood classification of remote sensed data

Satellite images are the best way to identify flood pretentious areas. Once we identify flood pretentious regions, then it is possible to identify the portion of vegetation area, residential area, water area, etc. But satellite images are very complex images from which data extraction is a very crucial task and it is also very difficult to identify pre-flood and post-flood images from large sets of data. So many techniques are used, but accuracy is still a major constraint. Thus, in this paper, the proposed nature-inspired algorithm is explained, which is inspired by the foraging technique of zebra animals and deep learning classification. Major focus on three phases of the proposed model: data processing, feature extraction and classification. Various comparison matrices are used to prove that the proposed algorithm is better than the existing algorithms.