Fully Dynamic Algorithms for Chamfer Distance
By: Gramoz Goranci , Shaofeng Jiang , Peter Kiss and more
We study the problem of computing Chamfer distance in the fully dynamic setting, where two set of points $A, B \subset \mathbb{R}^{d}$, each of size up to $n$, dynamically evolve through point insertions or deletions and the goal is to efficiently maintain an approximation to $\mathrm{dist}_{\mathrm{CH}}(A,B) = \sum_{a \in A} \min_{b \in B} \textrm{dist}(a,b)$, where $\textrm{dist}$ is a distance measure. Chamfer distance is a widely used dissimilarity metric for point clouds, with many practical applications that require repeated evaluation on dynamically changing datasets, e.g., when used as a loss function in machine learning. In this paper, we present the first dynamic algorithm for maintaining an approximation of the Chamfer distance under the $\ell_p$ norm for $p \in \{1,2 \}$. Our algorithm reduces to approximate nearest neighbor (ANN) search with little overhead. Plugging in standard ANN bounds, we obtain $(1+ε)$-approximation in $\tilde{O}(ε^{-d})$ update time and $O(1/ε)$-approximation in $\tilde{O}(d n^{ε^2} ε^{-4})$ update time. We evaluate our method on real-world datasets and demonstrate that it performs competitively against natural baselines.
Similar Papers
Even Faster Algorithm for the Chamfer Distance
Computational Geometry
Makes comparing shapes much faster for computers.
Dynamic Diameter in High-Dimensions against Adaptive Adversary and Beyond
Data Structures and Algorithms
Keeps data points organized even when they change.
Fully Dynamic Euclidean Bi-Chromatic Matching in Sublinear Update Time
Data Structures and Algorithms
Finds best pairings faster for data comparisons.