STUDY AND IMPLEMENTATION OF SOME FEATURE REDUCTION TECHNIQUES FOR HYPERSPECTRAL IMAGING

Abstract

Hyperspectral data can produce data of fine spectral resolution for the identification of materials from satellite observations as the spectral resolution. As the spectral resolution increases, the capability to detect more detailed classes may also increase. However, problem such as, curse of dimensionality and Hughes phenomena, may arise if too many spectral bands are simultaneously used in the classification process. Dimension reduction may, therefore, be necessary to reduce the size and redundancy in remote sensing data, which in turn reduces the cost and complexity of image processing. Some emerging feature reduction techniques in the field of remote sensing such as Segmented Principal Component Analysis and Wavelet Based Feature Extraction are used for feature reduction. The reduced dataset shall be in the classified to produce land use land cover map. The effectiveness of both the feature reduction techniques for dimensionality reduction are highlighted on the basis of classification accuracy and/or reduced number of features. The study has been conducted on two datasets, AVIRIS sensor data and Hyperion data.