Robust online appearance models for visual tracking

Abstract
We propose a framework for learning robust, adaptive,appearance models to be used for motion-based tracking ofnatural objects. The approach involves a mixture of stableimage structure, learned over long time courses, along with2-frame motion information and an outlier process. An onlineEM-algorithm is used to adapt the appearance modelparameters over time. An implementation of this approachis developed for an appearance model based on the filterresponses from a steerable pyramid. This...

This publication has 13 references indexed in Scilit: