Single-lens multi-ocular stereovision using prism

Abstract

A novel single-lens stereovision system using prism is introduced. An image captured by this system can be split into multiple different sub-images which can be considered as the images captured by a group of virtual cameras generated by the prism. Hence this system can simultaneously obtain multiple different views of the same scene using only one camera, and the disparities among the views can be exploited to perform depth recovery like a typical stereovision system. According to different numbers of virtual cameras generated, this system can be categorized into: binocular, trinocular and multi-ocular systems which have been discussed separately. Two different approaches of modeling and determining this system are provided: one is based on camera calibration technique and another is based on geometrical analysis of ray sketching. The latter approach is of the most interest because of its simpler implementation and good accuracy. Experiment shows both approaches are effective.