Stitch4D: Sparse Multi-Location 4D Urban Reconstruction via Spatio-Temporal Interpolation

2026-04-09

AI Summary

Existing 4D reconstruction methods struggle to recover intermediate regions in sparse, non-overlapping urban scenes, often producing temporal artifacts. To address this, this work proposes Stitch4D, the first 4D reconstruction framework specifically designed for sparse multi-view urban environments. Stitch4D explicitly enhances spatial coverage by generating bridging views through spatio-temporal interpolation, thereby preventing geometric collapse. It jointly optimizes real and synthesized observations within a unified coordinate system and enforces cross-location consistency constraints to ensure spatio-temporal coherence of dynamic content. Experiments on the newly introduced Urban Sparse 4D benchmark demonstrate that the proposed method significantly outperforms existing approaches, achieving notable improvements in both visual quality and geometric consistency.

Abstract

Dynamic urban environments are often captured by cameras placed at spatially separated locations with little or no view overlap. However, most existing 4D reconstruction methods assume densely overlapping views. When applied to such sparse observations, these methods fail to reconstruct intermediate regions and often introduce temporal artifacts. To address this practical yet underexplored sparse multi-location setting, we propose Stitch4D, a unified 4D reconstruction framework that explicitly compensates for missing spatial coverage in sparse observations. Stitch4D (i) synthesizes intermediate bridge views to densify spatial constraints and improve spatial coverage, and (ii) jointly optimizes real and synthesized observations within a unified coordinate frame under explicit inter-location consistency constraints. By restoring intermediate coverage before optimization, Stitch4D prevents geometric collapse and reconstructs coherent geometry and smooth scene dynamics even in sparsely observed environments. To evaluate this setting, we introduce Urban Sparse 4D (U-S4D), a CARLA-based benchmark designed to assess spatiotemporal alignment under sparse multi-location configurations. Experimental results on U-S4D show that Stitch4D surpasses representative 4D reconstruction baselines and achieves superior visual quality. These results indicate that recovering intermediate spatial coverage is essential for stable 4D reconstruction in sparse urban environments.

Problem

• 4D reconstruction
• sparse views
• multi-location
• urban environments
• spatio-temporal interpolation

Innovation

• 4D reconstruction
• sparse multi-location
• spatio-temporal interpolation
• view synthesis
• urban scene modeling