Abstract

Convolutional neural networks (CNNs) can adapt to more complex data, extract deeper characteristics from images, and achieve higher classification accuracy in remote sensing image scene classification and object detection compared to traditional shallow-model methods. However, directly applying common-structure CNNs to pixel-based remote sensing image classification will lead to boundary or outline distortions of the land cover and consumes enormous computation time in the image classification stage. To solve this problem, we propose a high-resolution remote sensing image classification method based on CNN and the restricted conditional random field algorithm (CNN-RCRF). CNN-RCRF adopts CNN superpixel classification instead of pixel-based classification and uses the restricted conditional random field algorithm (RCRF) to refine the superpixel result image into a pixel-based result. The proposed method not only takes advantage of the classification ability of CNNs but can also avoid boundary or outline distortions of the land cover and greatly reduce computation time in classifying images. The effectiveness of the proposed method is tested with two high-resolution remote sensing images, and the experimental results show that the CNN-RCRF outperforms the existing traditional methods in terms of overall accuracy, and CNN-RCRF’s computation time is much less than that of traditional pixel-based deep-model methods.

Highlights

  • With the rapid development of remote sensing technology, a large volume of high-resolution remote sensing images is available

  • A Convolutional neural networks (CNNs)’s input is a feature map set, while its output is a category label; applying this structure directly to pixel-based remote sensing image classification will lead to boundary and outline distortions of the land covers in the result image

  • To classify high-resolution remote sensing images more effectively, this paper proposes the CNN-restricted conditional random field algorithm (RCRF), which has two advantages

Read more

Summary

Introduction

With the rapid development of remote sensing technology, a large volume of high-resolution remote sensing images is available. In the common CNN structure, the input is a feature map set and the output is a category label Applying this structure to pixel-based remote sensing image classification will lead to boundary and outline distortions of the land covers in the result image [32]. The motivation of this paper is to fully utilize the CNN’s classification ability, avoiding traditional CNN drawbacks of boundary or outline distortions of land cover, reducing computational time, and achieving the goal of realizing “reasonable training sample set size- > acceptable model training time- > acceptable entire image classification time- > higher classification accuracy” in high-resolution remote sensing image classification. The effectiveness of the proposed method is tested with two high-resolution remote sensing images, and the experimental results show that the CNN-RCRF outperforms the existing traditional methods in terms of overall accuracy, and CNN-RCRF’s computation time is much less than that of traditional pixel-based deep-model methods

CNN and Image Classification
Fully Connected CRF
Algorithm Realization and Test Images
Comparison of Classification Results of Two Study Images
CNN fusion MLP
CNN-RCRF
Method
Comparison of Scale
Findings
Comparison of Computation Time
Conclusions
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call