Multisensor Multiresolution Data Fusion for Improvement in Classification

摘要: The rapid advancements in technology have facilitated easy availability of multisensor and multiresolution remote sensing data. Multisensor, data contain complementary information fusion such may result in application dependent significant which otherwise remain trapped within. present work aims at improving classification by fusing features coarse resolution hyperspectral (1 m) LWIR fine (20 cm) RGB map comprises eight classes. class names are Road, Trees, Red Roof, Grey Roof, Concrete Vegetation, bare Soil Unclassified. processing methodology for data comprises dimensionality reduction, resampling interpolation technique registering the two images at same spatial resolution, extraction to improve accuracy. In case resolution RGB data, vegetation index is computed classifying morphological building is calculated buildings. order extract textural features, occurrence co-occurence statistics considered and the will be extracted from all three bands RGB After extracting Support Vector Machine (SVMs) has been used training classification. To increase accuracy, post processing steps like removal any spurious noise as salt pepper done followed filtering process by majority voting within objects better object