Abstract: Satellite multispectral remote sensing imagery has been used over decades for feature extraction. The spectral classes of the imagery are finally translated into the different feature types in the image interpretation process (image processing). Presently, classification of all feature types is a manual process. Local and global climatic variability and change is inevitable which makes satellite imagery redundant in a short span of time. Due to the above stated reasons, we need an efficient and fast automatic feature extraction algorithm for better observing and organization of the resources of Earth. This paper proposes a technique to extract urban built-up, land/vegetation and water features from Enhanced Thematic Mapper Plus (ETM+) (Landsat 7) imagery. The study used three indices, Normalized Difference Water Index (NDWI), Normalized Difference Vegetation Index (NDVI), and Normalized Difference Built-up Index (NDBI) to represent three major features on Earth: built-up land, open water body, and vegetation, respectively. Consequently, the seven bands of an original Landsat 7 image were reduced into three thematic-oriented bands derived from above indices, which were combined to compose a new image. This reduced data correlation and redundancy between original multispectral bands, with overall accuracy ranging from 91.5 to 98.5 percent. All these feature extractions will be helpful in developing information systems on water bodies, vegetation areas a-lnd urban sprawling.
Keywords: Automation, remote sensing, Landsat, spectral index ratio, feature extraction.