HandyDepth: Example-based stereoscopic hand depth estimation using Eigen Leaf Node Features

This paper presents a data-driven method to estimate a high quality depth map of a hand from a stereoscopic camera input by introducing a novel regression framework. The method first computes disparity using a robust stereo matching technique. Then, it applies Random Forest (RF) to learn the mapping between the estimated, noisy disparity and actual depth given ground truth data. We introduce Eigen Leaf Node Features (ELNFs) that perform feature selection at the leaf node in each RF tree to identify features that are most discriminative for depth regression. Experimental results demonstrate the promise of the method to produce high quality depth images of a hand using an inexpensive stereo camera.