Spectral Characteristics Research of the Hyperspectral Image Based on the Correlation Matrix

摘要

The hyperspectral image is a sort of high spectral resolution remote sensing image, having strong relevance in the spectrum. Due to human activities and the apparatus' precision, hyperspectral image has complicated characteristics. Traditionally, we often use correlation coefficient h as reference to calculate the adjacent bands of the hyperspectral image. This paper introduces a comprehensive approach to statistical correlations between different bands, and set up a method for observing spectral correlation characteristics. The method uses the correlation matrix to generate the grayscale image of correlation coefficient and to replace correlation coefficient curve between adjacent bands, in order to handle the hyperspectral image reference in a better way.