Decoupled Geometric Parameterization and its Application in Deep Homography Estimation
This work addresses the need for more interpretable and computationally efficient homography estimation for computer vision applications, though it is incremental as it builds on existing parameterizations.
The paper tackles the problem of estimating planar homography in computer vision by introducing a novel geometric parameterization that decouples similarity and kernel transformations, enabling direct homography estimation without solving linear systems and achieving performance comparable to existing methods.
Planar homography, with eight degrees of freedom (DOFs), is fundamental in numerous computer vision tasks. While the positional offsets of four corners are widely adopted (especially in neural network predictions), this parameterization lacks geometric interpretability and typically requires solving a linear system to compute the homography matrix. This paper presents a novel geometric parameterization of homographies, leveraging the similarity-kernel-similarity (SKS) decomposition for projective transformations. Two independent sets of four geometric parameters are decoupled: one for a similarity transformation and the other for the kernel transformation. Additionally, the geometric interpretation linearly relating the four kernel transformation parameters to angular offsets is derived. Our proposed parameterization allows for direct homography estimation through matrix multiplication, eliminating the need for solving a linear system, and achieves performance comparable to the four-corner positional offsets in deep homography estimation.