Hyperspectral image classification method based on correction prototype learning

Abstract

The invention discloses a hyperspectral image classification method based on correction prototype learning, and the method comprises the steps: selecting a hyperspectral scene image, and randomly extracting a part of samples as a training set; randomly selecting a support set and a query set; building a deep network model, and calculating an initial class prototype in the learning measurement space; building a convolutional neural network with a residual block; training the deep network model; selecting a test data set, randomly selecting supervision samples from the test data set, and performing correction standardization processing on distribution of the supervision samples; and calculating the Euclidean distance between a test set sample and the test class prototype. According to the invention, the hyperspectral image is classified by using the hyperspectral classification method based on correction prototype learning, and the support set and the query set are selected, so that the use of a large number of labeled samples is avoided, and the sample labeling cost is saved. Compared with a traditional prototype network, the method obtains better classification precision, and has important application value in the aspects of hyperspectral image earth surface fine classification and the like.

Description

technical field [0001] The invention belongs to the technical field of hyperspectral image classification, and in particular relates to a hyperspectral image classification method based on correction prototype learning. Background technique [0002] Hyperspectral remote sensing realizes the integration of map and spectrum, contains rich spectral information and spatial information, and has a wide range of applications in precision agriculture, military reconnaissance, geological exploration and other fields. Hyperspectral image classification is a hotspot in hyperspectral image research, and with the in-depth exploration and application of deep learning, hyperspectral image classification technology has made great progress. However, limited labeled samples are an important reason restricting the development of hyperspectral image classification. Most of the existing network models and methods are based on manually labeled sample data, but this method is time-consuming, labo...

AI Technical Summary

Problems solved by technology

[0003] The invention provides a hyperspectral image classification method based on corrected prototype learning to overcome the fact that most of the existing network models and methods are based on manually labeled sample data, which is time-consuming, laborious and inefficient, and limited labeled samples restrict hyperspectral images. Technical Issues in Taxonomy Development

Images