Design, Modeling and Moving Object Detection of Omni-directional Vision System Applied in Autonomous Soccer Robot

摘要

The autonomous soccer robot with conventional vision cameras has limited fields of view which can be overcome by moving the robot, thus increasing the control difficulty. Omni-directional vision system uses a combination of lenses and mirrors placed in a carefully arranged configuration to capture a much wider field of view. This paper introduced study on omnidirectional vision system applied in autonomous soccer robot and its imaging model. Mirror parameters of omni-directional vision sensor were designed using condition and equation of single viewpoint constraint, and dimension of the omni-directional vision system was determined. A set of models including reflect and refraction model were built. The relation between image point and corresponding world point was determined that can offer essential model and algorithm for omni-directional image process. Then, moving object detected technology based on omni-directional vision system is introduced. By the low resolution characteristic of omni-directional correction image, the algorithm effectively deals with problems of the noise and shadow during the abstraction of the foreground. Their practicability is tested through the simulation and real image experiments.