A Hyperspectral Image Classification Method Based on Hybrid Metrics

Abstract

The invention discloses a hyperspectral image classification method based on mixed metrics. The algorithm judges the degree of abnormality of the object based on the neighborhood density of the object. For k lof Any positive integer, defining the k-th distance of x as the distance between x and an object o, denoted as dis_k lof (x), where the distance between the object x and the object o is recorded as d(x, o), first use the labeled data set to train the classifier, and use the classifier to classify the unlabeled samples; judge the unlabeled samples Confidence, adding high confidence unlabeled samples to the set of labeled samples; use k‑NN to select the k closest to the labeled sample knn Neighboring samples are not marked, and the spatial distance and LOF distance are introduced while calculating the spectral distance between samples. And the adaptive method is used to determine the input parameters of k-NN and LOF, which effectively improves the efficiency of the algorithm's parameter search and effectively improves the classification accuracy of the classification algorithm. The classification performance of the proposed algorithm on these datasets is better than that of similar algorithms.

Description

technical field [0001] The invention relates to an image classification method, in particular to a hyperspectral image classification method based on mixed metrics. Background technique [0002] Hyperspectral remote sensing technology originated in the early 1980s and was developed on the basis of multispectral remote sensing technology. Hyperspectral remote sensing can obtain approximately continuous spectral curves in the visible light, near-infrared, short-wave infrared, mid-infrared and other electromagnetic spectrum ranges through imaging spectrometers, and organically combine the spatial information representing the geometric position relationship of ground objects with the spectral information representing the characteristics of ground objects. Together, it makes it possible to extract the detailed information of ground features. With the improvement of the spectral resolution of the new imaging spectrometer, people's understanding of the spectral attribute character...

AI Technical Summary

Problems solved by technology

[0004] At present, a single measurement method is used for hyperspectral image classification, and the reference efficiency of the algorithm and the classification accuracy of the classification algorithm are low

Images