Learnable Motion Coherence for Correspondence Pruning
This work provides a more robust method for correspondence pruning, which is crucial for computer vision tasks like camera pose estimation, benefiting researchers and practitioners working with dynamic scenes and sparse correspondence data.
This paper addresses the challenge of distinguishing true from false correspondences in computer vision, especially in sparse and uneven distributions, by introducing the Laplacian Motion Coherence Network (LMCNet). LMCNet learns motion coherence properties, combining global and local coherence to robustly detect inlier correspondences, achieving superior performance in relative camera pose estimation and correspondence pruning in dynamic scenes.
Motion coherence is an important clue for distinguishing true correspondences from false ones. Modeling motion coherence on sparse putative correspondences is challenging due to their sparsity and uneven distributions. Existing works on motion coherence are sensitive to parameter settings and have difficulty in dealing with complex motion patterns. In this paper, we introduce a network called Laplacian Motion Coherence Network (LMCNet) to learn motion coherence property for correspondence pruning. We propose a novel formulation of fitting coherent motions with a smooth function on a graph of correspondences and show that this formulation allows a closed-form solution by graph Laplacian. This closed-form solution enables us to design a differentiable layer in a learning framework to capture global motion coherence from putative correspondences. The global motion coherence is further combined with local coherence extracted by another local layer to robustly detect inlier correspondences. Experiments demonstrate that LMCNet has superior performances to the state of the art in relative camera pose estimation and correspondences pruning of dynamic scenes.