Hyperspectral remote sensing image classification method combining quick region growth superpixel segmentation

Abstract

The invention discloses a hyperspectral remote sensing image classification method combined with quick region growth superpixel segmentation. According to the method, superpixels are generated througha region growing method; wherein a distance map is constructed, the generation of superpixels is accelerated; a super-pixel segmentation image and a pixel-based hyperspectral classification image aresubjected to texture adaptive fusion; according to the hyperspectral image classification method, the problem that the classification precision is not high due to the fact that superpixels serve as one unit in a traditional hyperspectral image classification method based on the superpixels is solved, the hyperspectral image classification precision is improved, and the hyperspectral image classification method has important practical significance in the aspects of environment management, crop monitoring, mineral mapping and the like.

Description

technical field [0001] The invention relates to the field of hyperspectral remote sensing image classification methods, in particular to a hyperspectral remote sensing image classification method combined with fast region growing superpixel segmentation. Background technique [0002] As a key technology of earth remote sensing, hyperspectral remote sensing image classification has attracted more and more people's attention. It is applied in many fields such as environmental monitoring, urban surveying and mapping, and precision agriculture. In recent decades, a large number of image classification techniques have been introduced using the spectral information of hyperspectral images. Typical methods include multinomial logistic regression, random forest, sparse representation, and support vector machines. As the spatial resolution of hyperspectral remote sensing images has been significantly improved, researchers have attempted to use spatial information to improve classific...

AI Technical Summary

Problems solved by technology

However, there are still two major challenges in superpixel-based hyperspectral image classification: (1) Most of the existing superpixel algorithms are for two-dimensional natural image processing. For three-dimensional hyperspectral images, the performance of superpixel algorithms (2) Traditional hyperspectral classification methods based on superpixels generally use each segmented superpixel as a unit for image classification, which may improve the efficiency of the algorithm, but usually reduces the accuracy of classification

Images