COMPARATIVE ANALYSIS OF LANDUSE LAND COVER BETWEEN OPTICAL AND FUSED IMAGE WITH SAR

摘要

Image fusion is a technique that integrates complementary information from multiple remote sensing images such that the fused image is more suitable for processing task and information extraction. Passive sensors are capable of sensing the reflected electromagnetic energy in the visible and infrared region while active sensors provide additional information using microwave region. This broad spectrum provides more information of earth surface as compared to optical data alone. This study compares the land cover classification results of optical imagery (Landsat-8) and fused imagery (Landsat-8 and Sentinel-1 VV polarized data). The image fusion was then performed using wavelet transformation technique. The data were classified into four classes namely water bodies, built-up area, vegetation cover, and barren land. Google Earth and Landsat imagery were used as a reference image for accuracy assessment. The fused image showed higher accuracy than optical image i.e. Kappa coefficient increased from 0.78 to 0.9 and overall accuracy increased from 89.4% to 92.7%. This study indicates that multi-source information i.e., image fusion can significantly improve the interpretation and accuracy of classification.