Abandoned object position prediction based on mobile robot

Abstract

Visual surveillance with dynamic background is one of the most active research topics in computer vision. Especially, researches for abandoned detection in the public place have been caused by preventing terror and accident. Previous researches for abandoned detection have been based on a static image and used an image difference. In this paper we present an algorithm for a mobile robot and we use optical flow for detecting a static object and distances from abandoned objects. For this purpose, we find abandoned objects and predict a position of objects using a mobile robot visual system with a single camera.