A decoupled geometric parameterization of homographies, based on the similarity-kernel-similarity decomposition, matches four-corner offset accuracy in deep homography estimation and replaces the DLT solver with matrix multiplication.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Decoupled Geometric Parameterization and its Application in Deep Homography Estimation
A decoupled geometric parameterization of homographies, based on the similarity-kernel-similarity decomposition, matches four-corner offset accuracy in deep homography estimation and replaces the DLT solver with matrix multiplication.