A feature extraction method of hyperspectral image based on fusion of spatial and spectral information

Abstract

The invention provides a hyperspectral image feature extraction method, comprising the following steps: sampling, obtaining pixel point class center; measuring the similarity of pixel points in space,and finding the nearest center point for each pixel point; constructing pixel locus matrix; constructing sample center trajectory matrix; obtaining Eigenvalues and corresponding eigenvectors by decomposing the sample center trajectory matrix, screening the eigenvalues and corresponding eigenvectors, reconstructing the eigenvalues and eigenvectors to obtain the trajectory matrix, and constructingthe transformation matrix , multiplying the transformation matrix by the pixel point trajectory matrix to obtain the reconstructed trajectory matrix; obtaining a new hyperspectral image according to the sequence relation of the reconstructed trajectory matrix. The invention provides the hyperspectral image feature extraction method, which combines spatial information to extract features of the hyperspectral image, so that the new image features have small differences in categories and large differences between categories, which is beneficial to the classification of the images and improves theclassification accuracy.

Description

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine

Login to view more

AI Technical Summary

Benefits of technology

This patented technology allows us to combine both space and spectrum data from different types of imagery (hyperspectrum) into one more powerful way by combining them together without compromising their original purpose or performance. It can help identify specific objects like people's faces better than traditional methods such as color cameras but also enhance object recognition capabilities.

Problems solved by technology

This patented technical problem addressed in this patents relates to improving the quality of 3 dimensional imagery that contains both grayscale values and color components due to factors like blurriness or lack of detail during analysis. These issues have been identified with traditional techniques based on PCA, EMD, Wavelets, and other tools commonly employed in conventional systems. However, these existing approaches may result in poorly performing results when analyzed at different levels within an object's surface layer.

Images