Multi-scale network analysis method based on hybrid multilayer perceptron

Abstract

The invention discloses a multi-scale network analysis method based on a hybrid multilayer perceptron. The method comprises an MSC block and a UMLP block; compared with the prior art, the high-efficiency hyperspectral classification method is realized, the model size is only 0.185 M, the method can be well suitable for industrial requirements, and the method can be well applied to the hyperspectral field, forest transition monitoring and timely early warning for forest disasters such as fire disasters and the like.

Description

technical field

[0001] The invention relates to the field of hyperspectral images, in particular to a multi-scale network analysis method based on a hybrid multi-layer perceptron. Background technique

[0002] A. Huertas et al. were the first to apply hyperspectral imagery to urban analysis. The boundary of urban buildings is extracted by geometric topology theory. JB.Adams et al. analyzed the spectral bands through mathematical theory and combined with other factors to analyze hyperspectral data. Generally speaking, the early hyperspectral image research mainly used the theoretical knowledge of various disciplines such as geometry to analyze shallow information. State-of-the-art methods always imply deep learning typified by Convolutional Neural Networks (CNN). In addition, it has achieved great achievements and outstanding performance in semantic segmentation, classification, object detection, etc. A new framework with contextual CNNs is described that explores local c...

Examples

Embodiment Construction

[0027] The specific embodiments of the present invention are described below so that those skilled in the art can understand the present invention, but it should be clear that the present invention is not limited to the scope of the specific embodiments. For those of ordinary skill in the art, as long as various changes Within the spirit and scope of the present invention defined and determined by the appended claims, these changes are obvious, and all inventions and creations using the concept of the present invention are included in the protection list.

[0028] The present invention mainly includes Numpy, Pandas, Tensor, etc. through the existing deep learning framework Pytorch and the corresponding programming library. Among them, Pytorch mainly uses deep learning models, including linear modules, convolution modules, and parameter penalty modules.

[0029] The implementation principle of the specific scheme is as follows:

[0030] We convert segmentation into a classific...

technical field

[0001] The invention relates to the field of hyperspectral images, in particular to a multi-scale network analysis method based on a hybrid multi-layer perceptron. Background technique

[0002] A. Huertas et al. were the first to apply hyperspectral imagery to urban analysis. The boundary of urban buildings is extracted by geometric topology theory. JB.Adams et al. analyzed the spectral bands through mathematical theory and combined with other factors to analyze hyperspectral data. Generally speaking, the early hyperspectral image research mainly used the theoretical knowledge of various disciplines such as geometry to analyze shallow information. State-of-the-art methods always imply deep learning typified by Convolutional Neural Networks (CNN). In addition, it has achieved great achievements and outstanding performance in semantic segmentation, classification, object detection, etc. A new framework with contextual CNNs is described that explores local c...

Examples

Embodiment Construction

[0027] The specific embodiments of the present invention are described below so that those skilled in the art can understand the present invention, but it should be clear that the present invention is not limited to the scope of the specific embodiments. For those of ordinary skill in the art, as long as various changes Within the spirit and scope of the present invention defined and determined by the appended claims, these changes are obvious, and all inventions and creations using the concept of the present invention are included in the protection list.

[0028] The present invention mainly includes Numpy, Pandas, Tensor, etc. through the existing deep learning framework Pytorch and the corresponding programming library. Among them, Pytorch mainly uses deep learning models, including linear modules, convolution modules, and parameter penalty modules.

[0029] The implementation principle of the specific scheme is as follows:

[0030] We convert segmentation into a classific...

Images