A Residual-Dyad Encoder Discriminator Network for Remote Sensing Image Matching


Numan Khurshid1
Mohbat1
Murtaza Taj1
Computer Vision and Graphics Lab1
Department of Computer Science
Lahore School of Management Sciences
Lahore, Pakistan
Visual Computing Lab2
Faculty of Science
University of Ontario Institute of Technology
2000 Simcoe St. N., Oshawa ON L1G 0C5

Abstract

We propose a new method for remote sensing image matching. The proposed method uses encoder subnetwork of an autoencoder pre-trained on GTCrossView data to construct image features. A discriminator network trained on University of California Merced Land Use/Land Cover dataset (LandUse) and High-resolution Satellite Scene dataset (SatScene) computes a match score between a pair of computed image features. We also propose a new network unit, called residual-dyad, and empirically demonstrate that networks that use residual-dyad units outperform those that do not. We compare our approach with both traditional and more recent learning-based schemes on LandUse and SatScene datasets, and the proposed method achieves state-of-the-art result in terms of mean average precision and ANMRR metrics. Specifically, our method achieves an overall improvement in performance of 11.26% and 22.41%, respectively, for LandUse and SatScene benchmark datasets.


Proposed residual dyad unit.
Proposed residual dyad unit.
Unsupervised Autoencoder Features: Image input from left (to encoder sub-network) and outputs to the right of (decoder) network. z is taken as the feature vector of the given image.
Unsupervised Autoencoder Features: Image input from left (to encoder sub-network) and outputs to the right of (decoder) network. z is taken as the feature vector of the given image.
Network architecture of the proposed ResDyadDML that takes features from ResDyadAE for an image pair and predicts the matching score. Residual-dyad block has been integrated to boost the performance of the network.
Network architecture of the proposed ResDyadDML that takes features from ResDyadAE for an image pair and predicts the matching score. Residual-dyad block has been integrated to boost the performance of the network.
Finding matching images in LandUse dataset.
Finding matching images in LandUse dataset.
Comparative results of class-wise mAP among supervised (VGG16, GoogleNet, and SatResNet-50) and our proposed approach.
Comparative results of class-wise mAP among supervised (VGG16, GoogleNet, and SatResNet-50) and our proposed approach.
The proposed model seems to construct deep features that may be useful in domains other than remote image sensing.
The proposed model seems to construct deep features that may be useful in domains other than remote image sensing.

Publication

For technical details please look at the following publications