A Spatially-Aware Crowdsensing Framework for High-Fidelity 3D View Synthesis

3D view synthesis reconstructs immersive 3D scenes from 2D images, but high-fidelity results require spatially diverse inputs that are often unavailable in practice. We propose a spatially-aware crowdsensing framework (SACF) that employs a reward-guided recruitment mechanism to selectively acquire geometrically complementary images. At its core, a spatially-aware reward map prioritizes viewpoints distant from previously collected images, ensuring scene completeness. The framework iteratively recruits participants, updates synthesis, and evaluates quality until convergence. Experiments on real-world datasets demonstrate that SACF achieves high-fidelity 3D view synthesis with minimal participant overhead. We also discuss future directions to improve robustness, economic sustainability, trust, and automation in practical deployments.