13-6 Detection and Classification of Moving objects in Outdoor Surveillance Image Sequence

Abstract

For outdoor image surveillance, it is important to detect moving objects exactly. This paper deals with their detection and classification, tolerant to change of the surroundings. We integrate two detection methods, Directional Spatio-temporal Filter and Motion Salience Operator, through handing of a data array of an object label that is assigned to its corresponding moving object. The label array is used for filling in holes of object's regions and classification of moving objects as well as the integration.