LocalTrans: A Multiscale Local Transformer Network for Cross-Resolution Homography Estimation

Shao, Ruizhi; Wu, Gaochang; Zhou, Yuemei; Fu, Ying; Liu, Yebin

doi:10.1109/iccv48922.2021.01462

Cited by 34 publications

(24 citation statements)

References 35 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The bottom right subscript, namely 1 or 2, denotes the computed result at the 1/4 × 1/4 resolution or the 1/2 × 1/2 resolution. The homography matrices H 1 and H 2 from the 2 scales are composed to produce the final homography estimation for the 2-scale IHN as in [14,28]. In Section 4, we show that IHN only needs 2 scales to achieve very high accuracy than the 3 or 4 scales IC-LK based methods [5,40].…”

Section: Multiscale Strategymentioning

confidence: 99%

“…Challenges in Homography Estimation. According to the recent works [14,25,28,39,40], there are mainly two challenges in homography estimation. The first challenge comes from the photometric inconsistency such as illumination change or modality variation.…”

Section: Related Workmentioning

confidence: 99%

“…Global Motion Aggregator. The residual homography is estimated by the global motion aggregator, with the homography matrix parameterized by the displacement vectors of the 4 corner points as in [6,14,28].…”

Section: Iterative Homography Estimatormentioning

confidence: 99%

“…Homography Updater. As in [6,14,28], we parameterize the homography matrix using the displacement vectors of the 4 corner points of an image, namely the displacement cube D. In iteration k, D is updated as…”

Section: Iterative Homography Estimatormentioning

confidence: 99%

“…We evaluate IHN on both common and challenging datasets. We test IHN for static scenes on common MSCOCO dataset [15] as in [5,6,14,28,40]. We also evaluate IHN on the cross-modal Google Earth and Google Map & Satellite datasets [40].…”

Section: Datasetsmentioning

confidence: 99%

See 4 more Smart Citations

Iterative Deep Homography Estimation

Cao¹,

Hu²,

Sheng³

et al. 2022

Preprint

View full text Add to dashboard Cite

We propose Iterative Homography Network, namely IHN, a new deep homography estimation architecture. Different from previous works that achieve iterative refinement by network cascading or untrainable IC-LK iterator, the iterator of IHN has tied weights and is completely trainable. IHN achieves state-of-the-art accuracy on several datasets including challenging scenes. We propose 2 versions of IHN: (1) IHN for static scenes, (2) IHN-mov for dynamic scenes with moving objects. Both versions can be arranged in 1-scale for efficiency or 2-scale for accuracy. We show that the basic 1-scale IHN already outperforms most of the existing methods. On a variety of datasets, the 2-scale IHN outperforms all competitors by a large gap. We introduce IHN-mov by producing an inlier mask to further improve the estimation accuracy of moving-objects scenes. We experimentally show that the iterative framework of IHN can achieve 95% error reduction while considerably saving network parameters. When processing sequential image pairs, IHN can achieve 32.7 fps, which is about 8× the speed of IC-LK iterator. Source code is available at https://github.com/imdumpl78/IHN .

show abstract