A Multispectral Image Classification Method Based on Deep Fusion Residual Nets

Abstract

The invention discloses a hyperspectral image classification method based on a deep fusion residual network, which overcomes the complexity and time-consuming design of artificial selection of multiple weak classifiers and integration methods in the prior art, the cumbersome calculation process and the semi-supervised training method. The classification results have the disadvantages of the phenomenon of the same object with different spectra and the phenomenon of different objects with the same spectrum. The steps that the present invention realizes are: (1) input multispectral image; (2) to the image of each wave band of each multispectral image, remove ground object normalization process; (3) obtain multispectral image matrix; ( 4) Acquire the data set; (5) Build the deep fusion residual network; (6) Train the deep fusion residual network; (7) Classify the test data set. The invention has the advantages of complete learning of multispectral image features, more concise and clear process, and more accurate classification effect, and can be used for classification of hyperspectral images.

Description

technical field [0001] The invention belongs to the technical field of image processing, and further relates to a multispectral image classification method based on a deep fusion residual network in the technical field of multispectral image classification. The invention can be used to classify ground objects including water areas, fields, ground objects and the like in multi-spectral images. Background technique [0002] Multispectral image is a kind of remote sensing image, which refers to the image formed by the reflection and transmission of electromagnetic waves in multiple bands, including visible light, infrared, ultraviolet, millimeter wave, X-ray, gamma-ray reflection or transmission image. As the basic research of multispectral images, multispectral image classification has always been an important means of information acquisition for multispectral images. Its main goal is to divide each pixel in the image into different categories according to the spatial geometri...

AI Technical Summary

Problems solved by technology

However, the disadvantage of this method is that it needs to select a variety of weak classifiers for ensemble learning, the selection of weak classifiers and the design of the ensemble method rely on human experience, and the training of multiple classifiers is complex and time-consuming. Time

The disadvantage of this method is that the calculation process of this method is cumbersome, and the semi-supervised clustering method is used, which leads to the phenomenon of different spectra of the same object and the same spectrum of different objects in the classification results, which affects the classification accuracy.

Images