Visual Tracking and Recognition Using Appearance-Adaptive Models in Particle Filters

DMCA

Visual Tracking and Recognition Using Appearance-Adaptive Models in Particle Filters (2004)

Cached

Download Links

Other Repositories/Bibliography

DBLP

by Shaohua Zhou , Rama Chellappa , Baback Moghaddam

Venue:	IEEE Transactions on Image Processing
Citations:	189 - 12 self

BibTeX

@ARTICLE{Zhou04visualtracking,
    author = {Shaohua Zhou and Rama Chellappa and Baback Moghaddam},
    title = {Visual Tracking and Recognition Using Appearance-Adaptive Models in Particle Filters},
    journal = {IEEE Transactions on Image Processing},
    year = {2004},
    volume = {13},
    pages = {1434--1456}
}

OpenURL

Abstract

We present an approach that incorporates appearance-adaptive models in a particle filter to realize robust visual tracking and recognition algorithms. Tracking needs modeling inter-frame motion and appearance changes whereas recognition needs modeling appearance changes between frames and gallery images. In conventional tracking algorithms, the appearance model is either fixed or rapidly changing, and the motion model is simply a random walk with fixed noise variance. Also, the number of particles is typically fixed. All these factors make the visual tracker unstable. To stabilize the tracker, we propose the following modifications: an observation model arising from an adaptive appearance model, an adaptive velocity motion model with adaptive noise variance, and an adaptive number of particles. The adaptivevelocity model is derived using a first-order linear predictor based on the appearance difference between the incoming observation and the previous particle configuration. Occlusion analysis is implemented using robust statistics. Experimental results on tracking visual objects in long outdoor and indoor video sequences demonstrate the effectiveness and robustness of our tracking algorithm. We then perform simultaneous tracking and recognition by embedding them in a particle filter. For recognition purposes, we model the appearance changes between frames and gallery images by constructing the intra- and extra-personal spaces. Accurate recognition is achieved when confronted by pose and view variations.

Keyphrases

particle filter appearance change recognition using appearance-adaptive model visual tracking gallery image appearance difference robust statistic motion model appearance-adaptive model following modification first-order linear predictor recognition algorithm accurate recognition robust visual tracking inter-frame motion occlusion analysis recognition purpose adaptive noise variance visual tracker previous particle configuration observation model long outdoor appearance model simultaneous tracking conventional tracking algorithm adaptive velocity motion model recognition need random walk adaptive appearance model extra-personal space adaptive number visual object indoor video sequence fixed noise variance adaptivevelocity model view variation experimental result

DMCA