Comparative study of feature extraction techniques for hyper spectral remote sensing image classification : A survey

Hyper spectral remote sensing image is also known as an “Imaging Spectrometry” is one of emerged technology for detection and identification of minerals, terrestrial vegetation, man-made materials, water bodies and backgrounds. The word “Hyper spectral” is used to discriminate sensors with many tens or hundreds of bands from the more traditional multiple sensors. The success of hyper spectral remote sensing image classification techniques is based on several factors where features have vital role. Different objects or materials reflect or absorb the sun's radiation in different ways. This is due to presence of variation in their surface features. For an object, the material, its physical, chemical state, the surface roughness and the geometric circumstances will influence the reflectance properties. The surface features like color, structure and surface texture are more useful in several applications. Extraction of above said features is essential step in order to correctly classify the objects. This paper gives brief comparison of several feature extraction approaches with its advantages and disadvantages.