Hyperspectral data classification method based on multi-layer convolution network and data organization and folding

Abstract

The invention relates to a hyperspectral data classification method based on a multi-layer convolution network and data organization and folding. The method comprises: step one, pretreatment is carried out before expanding and classification of three-dimensional hyperspectral data and a data matrix including effective spectrum information and a tag vector are obtained; step two, feature dimension expanding is carried out on the data matrix, and column-based folding and reorganization are carried out on a feature dimension to obtain a reorganized three-dimensional hyperspectral data input matrix; step three, a multi-layer convolution network structure parameter and an initial value are set; and step four, a feature and an error are calculated layer by layer by using forward propagation and BP algorithms, a network weight and a bias value are updated, iteration is carried out continuously to obtain a network stablity parameter, and then a network model for classification and a parameter for classification are obtained. Compared with other methods, the provided method has advantages of clear principle, clear structure, short identification time, and high detection identification rate; and the method being an effective classification method for hyperspectral data is suitable for rapid target detection and classification identification application of hyperspectral images.

Description

technical field [0001] The invention relates to a method for classifying hyperspectral pixel data, in particular to a method for classifying hyperspectral data based on a multi-layer convolutional network. Background technique [0002] Hyperspectral imaging is a method developed in recent years to obtain target imaging based on a large number of continuous narrow-band ground object spectral information. The spectral resolution of hyperspectral imaging is very high. Generally, the band interval is within 10nm. The target surface object is imaged with nanometer ultra-high spectral resolution, and hundreds of bands are acquired at the same time to form a continuous spectral image. Hyperspectral imaging technology is based on multispectral imaging, in the spectral range from ultraviolet to infrared, using imaging spectrometers to continuously image the target object with hundreds of spectral bands within the spectral coverage range, to obtain spatial feature information of the o...

AI Technical Summary

Problems solved by technology

[0005] At present, there are many classification techniques applied to hyperspectral image data. However, because the object is different from the traditional two-dimensional image, it has rich spectral information in the spectral dimension, so the processing complexity is greatly increased. The recognition results obtained by using different classification methods there is a big difference

Images