DynaMiTe: A Dynamic Local Motion Model with Temporal Constraints for Robust Real-Time Feature Matching

TL;DR

DynaMiTe is a lightweight, probabilistic feature matching pipeline that uses spatial-temporal cues and a dynamic local motion model to improve accuracy and efficiency in real-time visual odometry and SLAM.

Contribution

It introduces a descriptor-agnostic, probabilistic approach with a dynamic local motion model and temporal constraints for robust, real-time feature matching.

Findings

Outperforms state-of-the-art matching methods in accuracy

Achieves higher camera pose estimation precision

Operates at high frame rates with lower computational cost

Abstract

Feature based visual odometry and SLAM methods require accurate and fast correspondence matching between consecutive image frames for precise camera pose estimation in real-time. Current feature matching pipelines either rely solely on the descriptive capabilities of the feature extractor or need computationally complex optimization schemes. We present the lightweight pipeline DynaMiTe, which is agnostic to the descriptor input and leverages spatial-temporal cues with efficient statistical measures. The theoretical backbone of the method lies within a probabilistic formulation of feature matching and the respective study of physically motivated constraints. A dynamically adaptable local motion model encapsulates groups of features in an efficient data structure. Temporal constraints transfer information of the local motion model across time, thus additionally reducing the search space complexity for matching. DynaMiTe achieves superior results both in terms of matching accuracy and camera pose estimation with high frame rates, outperforming state-of-the-art matching methods while being computationally more efficient.