Abstract | ||
---|---|---|
We propose a new method for remote sensing image matching. The proposed method uses an encoder subnetwork of an autoencoder pretrained on the GTCrossView data to construct image features. A discriminator network trained on the University of California Merced land-use/land-cover data set (LandUse) and the high-resolution satellite scene data set (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 the LandUse and SatScene data sets, and the proposed method achieves the state-of-the-art result in terms of mean average precision and average normalized modified retrieval rank (ANMRR) metrics. Specifically, our method achieves an overall improvement in performance of 11.26% and 22.41%, respectively, for LandUse and SatScene benchmark data sets. |
Year | DOI | Venue |
---|---|---|
2020 | 10.1109/TGRS.2019.2951820 | IEEE Transactions on Geoscience and Remote Sensing |
Keywords | Field | DocType |
Feature extraction,Image matching,Measurement,Image retrieval,Training,Task analysis,Computer architecture | Residual,Computer vision,Discriminator,Image matching,Remote sensing,Artificial intelligence,Encoder,Dyad,Mathematics | Journal |
Volume | Issue | ISSN |
58 | 3 | 0196-2892 |
Citations | PageRank | References |
0 | 0.34 | 56 |
Authors | ||
4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Numan Khurshid | 1 | 2 | 1.72 |
Mohbat Tharani | 2 | 0 | 1.35 |
Murtaza Taj | 3 | 250 | 18.85 |
Faisal Z. Qureshi | 4 | 0 | 0.34 |