A hyperspectral image classification method for multi-feature class sub-dictionary learning

Abstract

The invention discloses a hyperspectral image classification method for multi-feature class sub-dictionary learning, which comprises the following steps: extracting a plurality of complementary feature data of the hyperspectral image; using a MFKCSDL model to learn the corresponding class sub-dictionary for each training sample; dividing hyperspectral images into several spatial groups by watershed-based image segmentation; applying the class sub-dictionary combinations to the MFKJSR model to obtain the multi-feature representation coefficients of pixel points in each spatial group; predictingclass labels of pixels in the spatial group by minimizing the multi-feature reconstruction error of all pixels in the spatial group. The invention can effectively improve the discrimination ability of the dictionary and further improve the classification accuracy of the hyperspectral image.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a hyperspectral image classification method for multi-feature class sub-dictionary learning. Background technique [0002] The problems of high dimensionality of hyperspectral image data, few training samples and high similarity between spectral bands have brought many challenges to its classification task. The classification model that only uses the spectral characteristics of a single pixel point is easily affected by factors such as "same object with different spectrum, same spectrum with different object", and the classification accuracy is low. With the deepening of research, researchers found that the spatial information describing the relationship between neighbors in hyperspectral images helps to further improve the classification accuracy, and then proposed a classification model that uses a large number of spectral information and spatial informatio...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the technical problems raised by the above-mentioned background technology, the present invention aims to provide a hyperspectral image classification method for multi-feature sub-dictionary learning, which solves the problems of the same object with different spectra and same spectrum with different objects in the hyperspectral image, and improves the high Spectral Image Classification Accuracy

Images