High-resolution remote sensing image classification method based on novel feature pyramid depth network

Abstract

The invention discloses a high-resolution remote sensing image classification method based on a novel feature pyramid depth network. The method comprises the following steps: firstly, designing a novel deep convolutional neural network on the basis of a ResNet34 network model; secondly, inputting the high-resolution remote sensing image into the network for training, and taking the output of eachmain convolution layer of ResNet34 as a subsequent input feature; fusing the input features by using a feature pyramid network to form new features; then, fusing the new deep-layer features and the new shallow-layer features to serve as inputs of an upper branch and a lower branch, and designing two residual blocks and a global average pooling layer on each branch; and fusing the features of the upper and lower branches and then sending to a full connection layer, and classifying the remote sensing images through a SoftMax layer. According to the invention, feature extraction and fusion are carried out on the high-resolution remote sensing image based on the deep learning theory, so that each feature is enhanced. The new features are fused again and then sent to the upper branch and the lower branch to learn image-level features, and experiments prove that the proposed method can achieve a good classification effect.

Description

technical field [0001] The invention belongs to the field of image processing, and in particular relates to a method for classifying high-resolution remote sensing images based on deep learning. Background technique [0002] Remote sensing generally refers to remote non-contact detection technology. Since different objects have obvious differences in the spectral effects of electromagnetic waves in the same band, remote sensing technology equipment analyzes the object's spectrum based on this principle, thereby realizing the recognition of distant objects. Common remote sensing technologies can be divided into multispectral, hyperspectral, and synthetic aperture radar, and the generated remote sensing images have different spatial, spectral, and temporal resolutions. Spatial resolution refers to the size or size of the smallest unit that can be distinguished in detail on remote sensing images. With the continuous development of remote sensing technology, the spatial resolu...

AI Technical Summary

Problems solved by technology

[0009] (2) The existing remote sensing image classification methods usually use bilinear interpolation to upsample the image. Although this method is outstanding in the traditional method with a small total sample size, the bilinear interpolation method has a large amount of calculation. Large, high-frequency components of the image are severely damaged after linear interpolation, and the edges of instance objects are blurred. It is not suitable for the modern deep learning field that requires high precision and rich sample data.

Images