Dimensionality Reduction Technique for Hyperspectral Remote Sensing Image Classification

Abstract

Hundreds of adjacent spectral bands are typically present in hyperspectral photographs. The problem with dimensionality arises from the high correlation and redundancy of such high-dimensional data. Prior to conducting additional analysis, such as classifying the land cover and identifying targets, these bands must be reduced. Remote sensing, seed viability analysis, biotechnology, environmental monitoring, food, pharmaceuticals, medical diagnosis, forensics, thin films, oil, and gas are just a few of the fields and technologies where the HSI is extensively used. The dimensionality issue can be solved and the size of the hyperspectral data is reduced effectively with feature extraction. So, the primary goal of the research is to use feature extraction and dimensionality reduction to lower the dimensions of the hyperspectral images that portray the image. This will facilitate the use of machinery to categorise various classes.