Parallel implementation of 3D object reconstruction in a robotic navigational environment

Stereo vision systems in a Robotic navigation environment determine the depth in the form of a depth map image from two or more images which are taken at the same time, but from slightly different viewpoints. Ground suppression from depth map is essential for object reconstruction in 3D environment as the ground surface is considered as traveling medium than as an object. Ground separated depth map needs to be segmented to identify the objects of interest. Objects identified below the threshold size need to be removed from the scene. In this paper we propose a Fuzzy-based parallel object clustering algorithm to reconstruct the objects of interest in 3D environment from the ground suppressed depth map. The depth map is divided into Fuzzy layers in X-Z plane and these Fuzzy layers are processed in parallel to reconstruct the objects of interest in 3D environment using region growing method on a cluster computing setup. Experimental results show that Fuzzy logic effectively handles the uncertainty in object reconstruction process. The parallel implementation of the algorithm drastically reduces the time required for object reconstruction process in congested navigational environments.