Method for carrying out tree species classification on hyperspectral data of artificial forest by using three-dimensional convolutional neural network

Abstract

The invention discloses a method for carrying out tree species classification on hyperspectral data of an artificial forest by using a three-dimensional convolutional neural network. The method is suitable for obtaining distribution information of artificial forest species, and belongs to the technical field of machine learning in forestry application. The key technical points comprise: 1, taking image data of a test area, and building a sample data set according to ground survey points; 2, building a three-dimensional convolutional neural network, and completing training of a model; and 3, predicting the image by using the trained model, and completng tree species distribution drawing and precision evaluation. The key problems to be solved comprise: 1, only a small number of samples are needed to complete the model training, and the workload of ground survey is reduced; and 2, compared with the traditional hyperspectral classification method, the method does not need to perform feature extraction and screening in advance, and is higher in classification precision. The method is suitable for hyperspectral data tree species classification, the three-dimensional convolutional neural network is applied to artificial forest tree species classification for the first time, and the result can provide basic data basis for forest resource investigation and tree species distribution information acquisition.

Description

[0001] 1. Technical field [0002] The invention relates to a method for classifying artificial forest hyperspectral images in the intersecting field of machine learning and forest resource investigation, in particular to a method for classifying hyperspectral remote sensing data of artificial forest areas using a deep convolutional neural network without requiring a large number of label samples. The classification method is applicable to the acquisition of plantation tree species distribution information, and belongs to the technical field of machine learning in forestry application. [0003] 2. Technical background [0004] Accurately obtaining forest tree species and spatial distribution information is of great significance for understanding forest ecosystem structure, function, succession, and biodiversity research, and is one of the most basic and key indicators in forest resource monitoring. [0005] Traditional tree species identification methods often rely on human fie...

AI Technical Summary

Problems solved by technology

[0008] In order to solve the problems of low precision and low efficiency in traditional tree species classification, the present invention aims to provide a three-dimensional deep learning network, which improves the traditional two-dimensional convolutional neural network, by adding spectral dimensions to the input of the neural network, giving The function of joint processing of spatial and spectral information by neural network

Images