Hyperspectral Remote Sensing Image Classification Method Based on Spatial Regularized Manifold Learning Algorithm

Abstract

The invention provides a hyperspectral remote sensing image dimensionality reduction and classification method based on a space regularized manifold learning algorithm, comprising the following steps: dividing the hyperspectral remote sensing image into a plurality of sub-blocks; and then randomly selecting some data points among them as connection data ; Combine the connection data and the data of each sub-block to obtain the enhanced sub-block data; calculate the graph Laplacian matrix corresponding to the LLE algorithm and the space regular constraint for each enhanced sub-block, and obtain a composite Laplacian matrix, The eigenvalue decomposition of the matrix is ​​performed to obtain the dimensionality reduction results; the dimensionality reduction results are aligned to obtain the dimensionality reduction results of the entire image; finally, the dimensionality reduction data is classified. The invention effectively combines data space information under the framework of a manifold learning algorithm, and adopts an image block and alignment strategy to maximize the effect of space regular constraints. The proposed algorithm shows good applicability to the classification of various hyperspectral remote sensing data, and can significantly improve the classification accuracy of hyperspectral remote sensing images.

Description

technical field [0001] The invention belongs to the technical field of remote sensing image processing, and specifically relates to a space-regularized manifold learning dimensionality reduction algorithm, combined with image block and dimensionality reduction result alignment technology, to maximize the role of space regular constraints, and to be used for hyperspectral remote sensing Image dimensionality reduction and classification. Background technique [0002] Hyperspectral remote sensing is a remote sensing science and technology with high spectral resolution. It has the characteristics of "integration of map and spectrum". Spectral and spatial information have important applications in the fine classification of ground objects. [0003] The amount of hyperspectral data is large, and there are problems of data redundancy and dimensionality disaster. Dimensionality reduction is an effective way to solve this problem. Dimensionality reduction can not only reduce the am...

AI Technical Summary

Problems solved by technology

Due to the high computational complexity of the semi-supervised classification algorithm, only a part of the data in the image can be selected to participate in the calculation. If the selected data is scattered in space, there are fewer other data points in the spatial neighborhood of a data point, and the spatial regularity constraints will be limited

Images