Hyperspectral image classification method based on texture features and affine propagation cluster algorithm

Abstract

The invention discloses a hyperspectral image classification method based on texture features and an affine propagation cluster algorithm, and mainly solves a problem that the classification effect ispoor in the prior art. The implementation scheme of the invention is that the method comprises the steps: 1), reading hyperspectral image data; 2), calculating a similarity matrix according to the texture features of bands; 3), carrying out the clustering of bands according to the similarity matrix through an AP algorithm, and obtaining band clustering results of the hyperspectral image; 4), calculating the optimal band of a band subset in each clustering result; 5), obtaining training set and test set sample coordinates of a calibration image and corresponding labels; 6), obtaining a training set and a test set at the optimal band according to the training set and test set sample coordinates; 7), training an SVM through the training set, and obtaining a training model; 8), predicting thetype of the test set according to the training model, and obtaining a classification result of the hyperspectral image. The method improves the classification precision of the hyperspectral image, and can be used for the recognition of remote sensing images of precision agriculture, ecological environment monitoring and urban planning.

Description

technical field [0001] The invention belongs to the technical field of remote sensing image processing, and in particular relates to a hyperspectral image classification method, which can be used to identify remote sensing images in agricultural precision, ecological environment monitoring and urban planning. Background technique [0002] Hyperspectral image processing has become one of the frontier technologies in the field of remote sensing. Hyperspectral remote sensing refers to the technology that imaging spectrometers acquire image data on tens to hundreds of very narrow and continuous spectral segments in the ultraviolet, visible, near-infrared, and mid-infrared regions of the electromagnetic spectrum. Retrieve relevant data from the target. On the one hand, hyperspectral data can distinguish these surface materials with sufficient spectral resolution, and on the other hand, because the reflected reflection spectra of ground objects can be compared with the measured v...

AI Technical Summary

Problems solved by technology

Since these methods are based on the similarity between the bands to select each type of representative bands to form a band subset, while ignoring the correlation between the bands of each type of cluster and the information content of the bands, resulting in poor performance of the selected bands, Image classification accuracy is low

Images