Deep neural network space spectrum classification method for high-spectral image

Abstract

The invention relates to a deep neural network space spectrum classification method for a high-spectral image and belongs to the technical field of deep learning and high-spectral remote sensing image classification. In the method, grouped space spectral features are used as input, according to input grouping features, a regularization item is added to an optimization target at a first layer of a deep neural network, and extraction of the space spectral features and waveband selection are realized. The method takes algorithm features of a deep belief network into consideration, also takes features of space information into consideration, performs individual processing on space groups of each waveband and is different from a deep convolutional network in which parameters in a convolutional nucleus are the same; and the algorithm can automatically attenuate weights of wavebands having quite small classification effects, realizes adaptive feature extraction and waveband selection, can obtain better classification accuracy compared to the typical deep belief network and has wide application prospect.

Description

technical field [0001] The invention belongs to the technical field of deep learning and hyperspectral remote sensing image classification, and relates to a deep neural network spatial spectrum classification method for hyperspectral images. Background technique [0002] With the continuous improvement of hyperspectral remote sensing image sensor technology, the spatial resolution and spectral resolution of hyperspectral remote sensing images have been greatly improved, whether on spaceborne or airborne, which makes the application of hyperspectral remote sensing images more and more widely. At the same time, with the improvement of spatial resolution and spectral resolution, it has also brought about a sharp increase in data dimensions, which can reach hundreds of dimensions. thousands of dimensions. This makes the previous algorithms that have good performance in low-dimensional space face great challenges in high-dimensional space. [0003] At present, the main methods...

AI Technical Summary

Problems solved by technology

[0005] Other shallow classifiers such as support vector machines, multi-classification logistic, etc., although the model is more concise, but the classification accuracy is not high enough

[0006] Existing deep learning algorithms such as the deep belief network do not consider the spatial information around the target pixel, while the 2-D deep convolutional network takes the spatial information into account, but its convolution kernel parameters are the same, which will undoubtedly bring loss of information

Images