Depth from Defocus via Active Quasi-random Point Projections

Abstract

Depth sensing has many practical applications in vision-related
tasks. While many different depth measurement techniques exist
and depth camera technologies are constantly being advanced, active
depth sensing still rely on specialized hardware that are highly
complex and costly. Motivated by this, we present a novel technique
for inferring depth measurements via depth from defocus using
active quasi-random point projection patterns. A quasi-random
point projection pattern is projected onto the scene of interest, and
each projection point in the image captured by a camera is analysed
using a calibration model to estimate the depth at that point.
The proposed method has a relatively simple setup, consisting of a
camera and a projector, and enables depth inference from a single
capture. Furthermore, the use of a quasi-random projection pattern
can allow us to leverage compressive sensing theory to produce
full depth maps in future applications. Experimental results
show the proposed system has strong potential for enabling active
depth sensing in a simple, efficient manner.