Classification for Remote Sensing Data With Improved CNN-SVM Method

Xiankun Sun,Chengfan Li,Wen Si,Lan Liu,Junjuan Zhao,Jingyuan Yin

doi:10.1109/access.2019.2952946

Xiankun Sun, Chengfan Li + Show 4 more

Open Access

PDF Available

https://doi.org/10.1109/access.2019.2952946

Copy DOI

Export

Save

Cite

Abstract
Highlights/Summary
Full-Text PDF
Similar Papers

Abstract

Listen

The efficient classification of remote sensing images (RSIs) has become the key of remote sensing application. To tackle the high computational cost in the traditional classification method, in this paper we propose a new RSI classification method based on improved convolutional neural network (CNN) and support vector machine (SVM) (CNN-SVM). In this method, we first designed a seven-layer CNN structure and took the ReLU function as the activation function. We then inputted the RSI into the CNN model and extracted feature maps and replaced the output layer of the CNN network via training the feature maps in the SVM classifier. Next, taking the simulation experiments of MNIST handwritten digital dataset and UC Merced Land Use remote sensing dataset as examples, we tested and verified the proposed method in this experiment. Finally, the empirical study of volcanic ash cloud (VAC) classification from moderate resolution imaging spectroradiometer (MODIS) RSI was carried out and evaluated. The experimental results show that compared with the traditional methods, the proposed method has lower loss value and better generalization in modeling training; the total classification accuracy of VAC and Kappa coefficient reached 93.5% and 0.8502, respectively, and achieved preferable VAC identification and visual effects. It will enhance the classification accuracy to the massive remote sensing data.

Highlights

Image classification is method of image processing that the different type of information is classified into the correspond categories in accordance with the feature reflected in the image [1], [2]
The classification method based on convolutional neural network (CNN) and support vector machine (SVM) model can overcome limitations of traditional methods and has become a major research of remote sensing images (RSIs) classification under the background of big data and deep learning
An improved CNN-SVM classification method from RSIs is proposed based on CNN and SVM classifier

Summary

INTRODUCTION

Image classification is method of image processing that the different type of information is classified into the correspond categories in accordance with the feature reflected in the image [1], [2]. Via constructing test datasets including different resolutions, sizes, sea conditions and sensor types, Li et al [28] proposed a new ship target detection method combined with feature aggregation and migration learning from SAR images. As it describes, the focus of CNN is mainly on the feature extraction and classifier design closely related to RSI types. The weight of convolutional kernel can be obtained by training, and the weight value does not change in the process of convolution calculation It greatly reduces the computational complexity of training parameters and the overfitting problem, and improves the generalization capability of CNN model to some extent. CNN network structure models contain LeNet, AleXNet, GoogLeNet and VGGNet, etc

SVM METHOD

Findings

CONCLUSION