Problem
Research questions and friction points this paper is trying to address.
Addresses event association and fusion for object tracking
Robustly selects true models from event data hypotheses
Handles high speed, motion blur, and dynamic range
Event Data Association via Robust Model Fitting for Event-based Object Tracking
🤖 AI Summary
Event cameras face significant challenges in associating asynchronous event streams under high-speed motion, motion blur, and high dynamic range conditions. Method: This paper proposes EDA (Entropy-driven Deterministic Association), a novel framework that introduces multi-structure geometric model fitting into event association. EDA fuses asynchronous event streams via entropy-driven weighting to generate deterministic trajectory hypotheses; it further incorporates a two-stage robust weighting scheme and an adaptive true-model-number estimation strategy to enable RANSAC-style robust multi-structure fitting and dynamic model selection. Contribution/Results: Experiments demonstrate that EDA substantially improves trajectory modeling accuracy and noise robustness. It achieves superior target tracking performance over state-of-the-art methods in challenging scenarios, particularly those involving rapid motion, severe motion blur, and extreme illumination variation.
📝 Abstract
Event-based approaches, which are based on bio-inspired asynchronous event cameras, have achieved promising performance on various computer vision tasks. However, the study of the fundamental event data association problem is still in its infancy. In this paper, we propose a novel Event Data Association (called EDA) approach to explicitly address the event association and fusion problem. The proposed EDA seeks for event trajectories that best fit the event data, in order to perform unifying data association and information fusion. In EDA, we first asynchronously fuse the event data based on its information entropy. Then, we introduce a deterministic model hypothesis generation strategy, which effectively generates model hypotheses from the fused events, to represent the corresponding event trajectories. After that, we present a two-stage weighting algorithm, which robustly weighs and selects true models from the generated model hypotheses, through multi-structural geometric model fitting. Meanwhile, we also propose an adaptive model selection strategy to automatically determine the number of the true models. Finally, we use the selected true models to associate and fuse the event data, without being affected by sensor noise and irrelevant structures. We evaluate the performance of the proposed EDA on the object tracking task. The experimental results show the effectiveness of EDA under challenging scenarios, such as high speed, motion blur, and high dynamic range conditions.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Asynchronous event data fusion via information entropy
Deterministic model hypothesis generation strategy
Two-stage weighting algorithm for model selection
🔎 Similar Papers
No similar papers found.
