Automatic convolution kernel size determining convolutional neural network-based hyperspectral image classification method

Abstract

The invention provides an automatic convolution kernel size determining convolutional neural network-based hyperspectral image classification method. The method comprises the following steps of: firstly clustering a plurality of image block samples with different sizes by using a K-means algorithm; designing a new evaluation index for a clustering result of the samples, the sizes of which are different and category quantities of which are uniformly distributed, and evaluating the clustering the result of the samples with different sizes by using the evaluation index, and determining a self-adaptive convolution kernel according to the evaluation result; and finally training a convolutional neural network of the self-adaptive convolution kernel, and classifying hyperspectral images by utilizing the trained network. Self-adaptive convolution kernels obtained by clustering and index evaluation are capable of representing data information more effectively, so that better hyperspectral imageclassification results can be obtained by utilizing the method.

Description

technical field [0001] The invention belongs to the technical field of hyperspectral image processing, and in particular relates to a hyperspectral image classification method based on a convolutional neural network for automatically determining the size of a convolution kernel. Background technique [0002] Currently, there are two main types of hyperspectral image pixel-level classification methods: one is based on artificially designed features; the other is based on deep learning features. The document "Ding C, Li Y, Xia Y, et al. Convolutional Neural Networks Based Hyperspectral Image Classification Method with AdaptiveKernels[J]. Remote Sensing, 2017, 9(6): 618." discloses a hyperspectral image based on deep learning In the image classification method, the convolution kernel can be obtained through pre-learning, and the number of adaptive convolution kernels can be obtained through the literature design method. This document proposes to use the improved clustering alg...

AI Technical Summary

Problems solved by technology

[0003] In order to overcome the problem that the size of the convolution kernel needs to be manually set in the existing convolution neural network structure based on convolution kernel learning in advance and cannot adaptively represent the characteristics of data information, the present invention discloses a convolution algorithm based on automatically determining the size of the convolution kernel. The hyperspectral image classification method of the convolutional neural network introduces the K-means clustering algorithm into the convolution kernel learning of the convolutional neural network, that is, first uses the K-means algorithm to cluster multiple groups of samples of different sizes , and then, design a new evaluation index for the clustering results of samples of different sizes and non-uniform distribution of the number of categories, evaluate the clustering results of samples of different sizes, determine the adaptive convolution kernel according to the evaluation results, and finally, the The learned convolution kernel is introduced into the convolutional neural network and the fully connected layer of the network is trained, and the trained network is used to classify hyperspectral images

Images