In remote sensing images, extraction of high level features from flooded image captures the attention recent investigators. Moreover, higher level features attainment may lacks in high dimensionality and global representation. Here, thresholding and feature scaling are used for high level feature extraction. With a given input image dataset, image classification is performed with three diverse phases to extract features. As well, spatial index and scaling factors plays an essential role in this work. The results were validated using statistical measures like detection rate and false alarm rate on available datasets. Experiments on remote sensing image datasets demonstrate that the proposed mapping level shows higher progression towards classification performance. However, feature mapping outperforms the existing features extraction techniques from remote sensing images. Simulation is done with MATLAB environment and mapping of higher level features are extracted with level set analysis.