DSER: Spectral Epipolar Representation for Efficient Light Field Depth Estimation

TL;DR

The paper introduces DSER, a spectral epipolar regularization framework for efficient, accurate, and robust light field depth estimation, addressing challenges like occlusion and textureless regions.

Contribution

It proposes a novel spectral regularization approach in the epipolar domain combined with a hybrid inference pipeline for improved depth estimation.

Findings

DSER outperforms classical and hybrid baselines in accuracy and efficiency.

It produces more structurally consistent depth maps.

Spectral epipolar regularization proves effective for scalable light field depth estimation.

Abstract

Dense light field depth estimation remains challenging due to sparse angular sampling, occlusion boundaries, textureless regions, and the cost of exhaustive multi-view matching. We propose \emph{Deep Spectral Epipolar Representation} (DSER), a geometry-aware framework that introduces spectral regularization in the epipolar domain for dense disparity reconstruction. DSER models frequency-consistent EPI structure to constrain correspondence estimation and couples this prior with a hybrid inference pipeline that combines least squares gradient initialization, plane-sweeping cost aggregation, and multiscale EPI refinement. An occlusion-aware directed random walk further propagates reliable disparity along edge-consistent paths, improving boundary sharpness and weak-texture stability. Experiments on benchmark and real-world light field datasets show that DSER achieves a strong accuracy-efficiency trade-off, producing more structurally consistent depth maps than representative classical and hybrid baselines. These results establish spectral epipolar regularization as an effective inductive bias for scalable and noise-robust light field depth estimation.