Multi-scale collaborative representation hyperspectral classification method based on local adaptive dictionary

Abstract

The invention discloses a multi-scale collaborative representation hyperspectral classification method based on a local adaptive dictionary. The method comprises the steps that three-dimensional hyperspectral high-dimensional data is read; multiple scales of neighborhoods are determined; a combined signal matrix is constructed according to the hyperspectral data; the local adaptive dictionary is constructed according to the hyperspectral data; a corresponding coefficient matrix is solved; samples are reconstructed, and corresponding residual errors are calculated; residual error information corresponding to the neighborhoods of different scales is calculated; the residual errors of the neighborhoods of multiple scales are fused; and a hyperspectral pixel category is determined, and a classification result is obtained. The method has the advantages that spatial information in an image is fully utilized through the neighborhoods of multiple scales, the local adaptive dictionary is used to avoid irrelevant information, the classification image is good in visual effect, and classification precision is improved.

Description

technical field [0001] The invention relates to the technical field of remote sensing information processing, in particular to a multi-scale collaborative expression hyperspectral classification method based on a local adaptive dictionary. Background technique [0002] Hyperspectral images contain hundreds or thousands of band information, which can reflect the spectral characteristics of ground objects. At the same time, hyperspectral images are characterized by large data volume, high redundancy, and high dimensionality, and there is a strong correlation between bands. These characteristics bring challenges to subsequent processing. Hyperspectral image classification is mainly based on the different electromagnetic energy reflected by different ground objects, so as to show the difference of spectrum to realize the discrimination of different ground objects. The goal is to divide each pixel in the image into a category. With the development of technology, sparse represent...

AI Technical Summary

Problems solved by technology

[0003] The existing hyperspectral image classification methods mainly have the following problems: 1. The neighborhood information in the hyperspectral image is not fully utilized, and the neighborhood information of different scales has not been comprehensively considered

2. During the classification process, irrelevant information has not been effectively eliminated for specific pixels

[0007] 1. Neighborhood information in hyperspectral images is not fully utilized, and neighborhood information at different scales has not been comprehensively considered

[0008] 2. During the classification process, irrelevant information has not been effectively eliminated for specific pixels

[0009] The above problems lead to the inability of the hyperspectral image to be well expressed, resulting in low classification accuracy

Images