Hyperspectral image classification method based on image regular low-rank expression dimensionality reduction

Abstract

The invention discloses a hyperspectral image classification method based on image regular low-rank expression dimensionality reduction. The method includes the steps that a mean shift technology is used for conducting pre-segmentation on a hyperspectral image first, image regular low-rank coefficient expression is conduced on the hyperspectral image after pre-segmentation to obtain an image regular low-rank coefficient matrix, a characteristic value equation is constructed, a mapping matrix of the dimensionality reduction is studied, and original high dimensional data are transformed to low-dimensional space to be further classified. According to the hyperspectral image classification method, a hyperspectral image local manifold structure is excavated, the spatial distribution character of an original image is kept, effective dimensionality reduction space is studied, the classification accuracy of hyperspectral images is improved, computation complexity is lowered, the problems that the dimensionality of the hyperspectral image is too high so that the calculation amount can be large, and an existing method is low in classification accuracy are mainly solved, and the hyperspectral image classification method can be used for important fields such as precision agriculture, object identification and environment monitor.

Description

technical field [0001] The invention belongs to the technical field of remote sensing image processing, and relates to a low-rank representation method based on graph regularization, specifically a hyperspectral image classification method based on graph regularization and low-rank representation dimension reduction, which is used to solve hyperspectral remote sensing data dimension reduction Methods for Jane and Classification Problems. Background technique [0002] Hyperspectral remote sensing images use imaging spectrometers to simultaneously image surface objects with dozens or even hundreds of bands to form a three-dimensional data cube composed of continuous-band images to achieve simultaneous acquisition of spatial information, radiation information, and spectral information of ground objects. The characteristics of "integration" have improved the classification and monitoring capabilities of ground objects, and have been widely used in many military and civilian fiel...

AI Technical Summary

Problems solved by technology

Linear Discriminant Analysis (LDA) is a supervised linear dimensionality reduction method. Its main goal is to maximize the inter-class divergence while minimizing the intra-class divergence. LDA requires sufficient labeled samples, but high Obtaining sample labels for spectral images is very difficult, and the global linearity of PCA and LDA limits their effectiveness on non-Gaussian data

Nonlinear methods mainly include kernel-based methods and manifold learning methods in recent years, such as isometric feature mapping (ISOMAP), local manifold embedding (Locally linear embedding, LLE), etc. ISOMAP is a method that maintains The side-to-ground distance between two points on the manifold is used to maintain the global geometric characteristics of the data, which ensures the robustness and global optimality of the dimension reduction results, but the computational complexity is very high, and the memory requirements for computing devices are large.

[0004] The main problems of the current hyperspectral image dimensionality reduction methods are: the amount of data to be processed is large, and all the useful information provided by the hyperspectral image data cannot be well preserved; the dimensionality reduction method of manifold learning mines the local hyperspectral image Manifold structure information, but the computational complexity is high; the noise effect of the hyperspectral image is not considered in the process of dimensionality reduction

Images