Hyperspectral remote sensing image recognition method based on deep forest transfer learning

Abstract

The invention discloses a hyperspectral remote sensing image recognition method based on deep forest transfer learning. The method specifically comprises the following steps: S1, selecting a source domain data set and a target domain data set; and S2, training the source domain data set by using ResNet-CNN, and extracting spectral information and spatial information. The invention relates to the technical field of remote sensing images. According to the hyperspectral remote sensing image recognition method based on deep forest transfer learning, through deep forest transfer learning, the method can be effectively suitable for the situation that training samples are insufficient, meanwhile, good model universality is achieved, the problems that the equipment requirement is high, and time is consumed in classification are solved, a double-branch network structure, namely ResNet-CNN is adopted as a feature extraction algorithm, and the recognition efficiency is improved. According to the method, the desired sample features can be efficiently extracted, the deep forest algorithm is adopted as a classifier, the classification precision is improved, the phenomena of parameter adjustment difficulty, overfitting and the like are avoided, the classification efficiency is finally improved, and the classification cost and the demand for training samples are reduced.

Description

technical field [0001] The invention relates to the technical field of remote sensing images, in particular to a hyperspectral remote sensing image recognition method based on deep forest transfer learning. Background technique [0002] Remote sensing is a long-distance, non-contact target detection technology, and it is an important means for people to study the characteristics of ground objects. It gradually develops to the residential band imaging, and at the same time presents the characteristics of high spatial resolution, high spectral resolution, high temporal resolution, etc., and hyperspectral remote sensing appears. The images obtained by hyperspectral remote sensing include rich spectral information of objects and objects. At the same time, it also has rich spatial context information, which provides sufficient feature information for the classification and identification of ground objects. means, traditional algorithms such as support vector machine, random fore...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the deficiencies of the prior art, the present invention provides a hyperspectral remote sensing image recognition method based on deep forest transfer learning, which solves the problem that the currently commonly used hyperspectral remote sensing image analysis algorithm contains a large number of hyperparameters that need to be adjusted, and the parameter adjustment process is cumbersome. The process of artificial parameter adjustment is difficult and slow, it is difficult to construct a general model, and the problem of high requirements for computing equipment

Images