Distance estimation using a panoramic sensor based on a novel spherical optical flow algorithm

To implement an unmanned autonomous obstacle avoidance system, this paper presents a novel distance estimation method based on an improved spherical optical flow algorithm using a panoramic sensor. The method uses a panoramic sensor to extend the field of view. This paper investigates a panoramic sensor imaging model and the optical flow field and then derives the improved spherical optical flow algorithm. Based on the spatial structure characteristics of the panoramic image, the proposed spherical optical flow smoothness constraint is introduced into the spherical optical flow field. Combing the smoothness constraint with the spherical optical flow equation, the spherical optical flow value is solved for. The motion model of a panoramic sensor, which is the distribution of spherical optical flow field with the condition that the panoramic sensor has self-translational motion, is developed. According to the spherical optical flow value, the motion model and the velocity of the panoramic sensor, a distance estimation mechanism and a computational method are proposed for an Unmanned Aerial Vehicle (UAV). Simulations show that the method can extract the distance information effectively and offers good accuracy. (c) 2012 Elsevier Ltd. All rights reserved.