Hyperspectral image classification method based on spatial information enhancement and deep belief network

Abstract

The invention discloses a hyperspectral image classification method based on spatial information enhancement and a deep belief network and mainly solves a problem that the accuracy improvement of the hyperspectral image classification in the existing technology is not obvious. The technical scheme is as follows: the method comprises the following steps of 1) carrying out normalization and band selection on an original hyperspectral image to obtain a hyperspectral image with a reflectance value of 0 to 1; 2) carrying out spatial information enhancement on the hyperspectral image through band grouping and ground and guided filtering; 3) constructing a deep belief network model according to the features of the hyperspectral image after spatial information enhancement; and 4) carrying out model training on the hyperspectral image after spatial information enhancement and utilizing the obtained model to carry out category prediction in order to obtain a classification result. According to the method, the spatial information of the hyperspectral image is effectively enhanced without losing the original spectral information, the classification accuracy is remarkably improved, and the method can be used for environment, climate, agriculture and mineral exploration.

Description

technical field [0001] The invention belongs to the fields of hyperspectral image processing and computer vision, and in particular relates to a hyperspectral image classification method, which can be used for environment, climate, agriculture and mineral detection. Background technique [0002] With the rapid development of spectral imaging technology and remote sensing technology, hyperspectral remote sensing technology has become a cutting-edge technology. Hyperspectral remote sensing images have high spectral resolution and high spatial resolution, which can achieve accurate classification of surface objects, so they can be used in many fields: such as environment, climate, agriculture, and mineral detection. [0003] At present, the main method to solve the problem of hyperspectral image classification is to extract spectral features and then classify with the help of classifiers. This method mainly focuses on traditional machine learning algorithms, among which the alg...

AI Technical Summary

Problems solved by technology

[0005] At present, the method of applying deep belief network to hyperspectral image classification tasks often extracts all pixels in the neighborhood of the pixel when introducing spatial information, and directly connects these pixels to form a spatial information vector, and then combines the spatial information vector with the spectrum The information vectors are concatenated into feature vectors that contain both spatial information and spectral information. There are two disadvantages of this operation. One is that neighborhood splicing will cause spectral confusion and affect the original spectral information; the other is that only the original spatial information of the hyperspectral image is used. , does not enhance the spatial information for the hyperspectral image characteristics, so the hyperspectral image classification algorithm based on the deep belief network based on this kind of spatial information introduction method has no obvious improvement in classification accuracy compared with traditional shallow network classification methods such as SVM

Images