Stereovision for obstacle detection on smart mobile devices: First results

摘要

Recent market deployment of smart mobile devices which feature synchronous stereo image acquisition has raised the question whether such devices can be used for real-time 3D environment reconstruction by stereovision. In this paper, we propose a stereovision approach that can run in real-time on smart mobile devices, and we evaluate its potential for developing driving assistance functions. The stereo approach is a sparse approach: edges are detected in the left image, correspondent right image points are determined using area-based matching, and each left-right pair of points is mapped in 3D by triangulation. Our experiments have proved that despite the limitations imposed by the mobile device, both reconstruction accuracy at short-medium distances and real-time processing can be achieved. Thus, developing driving assistance functions with such devices is possible for low vehicle speeds / short range scenarios, which often occur in urban environments.