Depth estimation via shift-photography in Alvarez Lens imaging systems

Abstract

The Alvarez lens imaging system employs a novel zooming and focusing mechanism achieved through the lateral displacement of its Alvarez lenses. To enhance the usability of the Alvarez lens, we propose a method that leverages the functionality of shift photography to perform depth estimation using multiple image inputs. We simulate the imaging perspectives and scene displacements during shift photography using the Unity 3D engine while simultaneously acquiring pixel-to-pixel depth information. Based on the point spread function (PSF) of the optical system, which varies with the field of view and depth, we can accurately simulate the imaging effects of the Alvarez lens system. Experimental results demonstrate that our approach achieves superior accuracy compared to monocular depth estimation.