Acquisition of sharp depth map from multiple cameras

Abstract

We present a method to estimate a dense and sharp depth map using multiple cameras. A key issue for obtaining sharp depth map is how to overcome the harmful influence of occlusion. Thus, we first propose an occlusion-overcoming strategy which selectively uses the depth information from multiple cameras. With a simple selection technique, we resolve the occlusion problem considerably at a slight sacrifice of noise tolerance. Another key issue in area-based stereo matching is the size of matching window. We propose to use a hierarchical estimation scheme that attempts to acquire a sharp depth map such that edges of the depth map coincide with object boundaries on the one hand, and reduce noisy estimates due to insufficient size of matching window on the other hand. Owing to the unique property of occlusion-overcoming strategy, we can utilize full benefit of hierarchical schemes. We show the method can produce a sharp and correct depth map for a variety of images.