Omnidirectional Vision for Mobile Robot Navigation

Abstract

Machine vision has been widely studied, leading to the discovery of many image-processing and identification techniques. Together with this, rapid advances in computer processing speed have triggered a growing need for vision sensor data and faster robot response. In considering omnidirectional camera use in machine vision, we have studied omnidirectional image features in depth to determine correlation between parameters and ways to flatten 3-dimensional images into 2 dimensions. We also discuss ways to process omnidirectional images based on their individual features.