Study On Object Extraction Based On Feature From Hyperspectral Image

In the recent decades,hyperspectral imaging technology has more and more broad prospects with developed rapidly,especially in remote sensing.This paper has proposed a feature extraction method for the data processing of hyperspectral image,which combines image fusion and decision tree algorithm.Firstly,the common image is used to explore the target feature extraction method.The effect of image fusion by the Gram-Schmidt method is better than that of using the principal component analysis algorithm for the common image selected.And then,the fusion image with better effect was select to conduct image registration.In combination with decision tree classification method,the spectral information of mountains,rivers and vegetation in the image were extracted successfully.According to the existing conditions,two groups of experimental samples were used to verify this method.One of groups that only use of principal component analysis can target the vast majority of information extraction effectively.For the second one,the samples were extracted by combining the principal component analysis with the decision tree classification,and so the target was successfully separated.The experimental results concluded that the proposed algorithm can separate and extract the target effectively,which can be applied to the feature extraction of hyperspectral image.Combining the image fusion algorithm with the decision tree algorithm have made some contributions for the further processing of the image.