Remote sensing image classification method based on Gabor features and EMAP features

Abstract

The invention discloses a remote sensing image classification method based on Gabor features and EMAP features. The method comprises the steps: extracting Gabor features from a remote sensing image, constructing a Gabor feature kernel, extracting EMAP features from the remote sensing image, and constructing an EMAP feature kernel; fusing the constructed Gabor feature kernel and EMAP feature kernel by using a feature composite kernel framework to obtain a feature composite kernel; and on the basis of the obtained feature composite kernel, classifying samples in the remote sensing image through a multi-term logistic regression model. According to the method, the Gabor features and the EMAP features are combined with spectral information and spatial information, so that the adverse effect of original spectral data is avoided, the problems of unbalanced categories and few samples are solved, the problem of information redundancy caused by relatively strong correlation of the spectral information is effectively solved, and the classification precision is improved.

Description

technical field [0001] The invention belongs to the field of interpretation, recognition and processing of remote sensing images, in particular to a classification method of hyperspectral remote sensing images based on Gabor features and extended multi-attribute profile (EMAP) features. Background technique [0002] In hyperspectral remote sensing images, the spectral information of different types of ground objects is different. Hyperspectral image classification technology is of great significance to promote the development of remote sensing technology applications, but this technology faces problems such as high dimensionality of hyperspectral image data, few labeled samples, mixed pixels, and low data quality. [0003] Aiming at the problem of few labeled samples of hyperspectral images, scholars at home and abroad have proposed a series of hyperspectral image classification algorithms. At present, the most widely used supervised learning algorithm. The supervised lear...

AI Technical Summary

Problems solved by technology

[0006] To this end, the present invention proposes a fusion feature expression method, which combines spectral information and spatial information by fusing Gabor features and EMAP features to avoid the adverse effects of original spectral data, overcome the problems of unbalanced categories and small number of samples, and effectively solve the problem of Due to the information redundancy problem caused by the strong correlation of spectral information, the classification accuracy is improved

Images