D^3S Consensus slashes baseline 10x with defocus-stereo depth fusion
4mm baseline achieves 1cm depth error up to 1.64m — smaller than any commercial stereo camera.
D^3S Consensus is a physics-based, closed-form algorithm that unifies depth-from-defocus (DfD) and stereo for highly accurate depth estimation beyond the depth-of-field. Using a pair of dual-defocus stereo images, it estimates an overdetermined set of depth predictions with a novel DfD theory—Dual Differential Defocus (D^3)—and stereo, then picks the most confident depth by enforcing consensus between these independent cues to reject unreliable estimates. D^3S achieves a comparable working range with 10x smaller baseline than previous triangulation-based systems, enabling compact passive binocular rangefinders with much smaller form factors. A prototype with only 4 mm baseline and 12 mm EFL generates up to 900×1800-pixel depth maps with 1-cm mean absolute error over 0.3–1.64 m from a snapshot, surpassing the reported accuracy of certain commercially available stereo cameras with much larger form factors.
- Closed-form algorithm fusing depth-from-defocus (DfD) and stereo cues for robust depth estimation
- Prototype uses only 4mm baseline and 12mm EFL, 10x smaller than conventional triangulation systems
- Achieves 1cm mean absolute error on 900×1800 depth maps over 0.3–1.64m range in a single shot
Why It Matters
Enables ultra-compact, passive depth sensors for smartphones, AR headsets, and drones without bulky baselines.