HMM-based Land Cover Information Extraction from Remote Sensing Imagery

Hidden Markov Model (HMM) offers significant potential for analyzing complex information that cannot be directly deciphered by humans. Two properties are essential for a Hidden Markov Model: classifying the desired hidden states from a chaotic observation sequence along with predicting missing data based on transfer probabilities, which are applied to two remote sensing tasks in this study: water body segmentation using the Normalized Difference Water Index (NDWI) and forest land prediction based on time-series data of the Normalized Difference Vegetation Index (NDVI). The water body segmentation task achieved an overall accuracy of 83.16%, while the forest land prediction task attained an accuracy of 87.78%. These results suggest that HMM holds considerable promise for advancing remote sensing image interpretation.